Управление ближним электромагнитным полем в резонансных наноструктурах золото-кремний тема диссертации и автореферата по ВАК РФ 01.04.05, кандидат наук Сунь Яли

• Сунь Яли
• кандидат наук
• 2021

Оглавление диссертации кандидат наук Сунь Яли

Реферат

Synopsis

Introduction

CHAPTER 1. Optical near-field studies

1.1 Theoretical near-field optics and detection

1.2 Near-field characterization and applications

1.3 Approaches to control near-field in nanoscale

1.4 Hybrid metal/dielectric systems

1.5 Chapter conclusions

CHAPTER 2. Experimental techniques and numerical methods

2.1 Fabrication methods

2.1.1 Lithography fabrication

2.1.2 Femtosecond laser treatment

2.2 Experimental characterization methods

2.2.1 Dark-field scattering

2.2.2 Raman scattering

2.2.3 Scanning near-field optical microscopy

2.2.4 Second harmonic generation

2.3 Numerical simulations

2.4 Chapter conclusions

CHAPTER 3. Field confinement in gold-silicon nanoantennas

3.1 Strong field localization in gold-silicon nanoantenna

3.2 Field enhanced Purcell factor

3.3 Influence of E-field enhancement on SHG

3.4 Chapter conclusions

CHAPTER 4. Control of optical field distribution in the near-field zone

through shape configuration of gold-silicon nanoantenna

4.1 Shape induced control of near fields in a single nanoantenna

4.2 Near field control in dimers of gold-silicon nanoantennas

4.3 Chapter conclusions

CHAPTER 5. Reconfiguration of near-field distribution in gold-silicon

heptamer

5.1 Studies of laser-induced modification of gold-silicon heptamer

5.2 Realization of near-field reconfiguration in hybrid heptamer

5.3 Far-field reconfiguration studies

5.4 Chapter conclusions

Conclusion

List of acronyms and symbols

Bibliography

APPENDIX A. Main journal papers

Введение диссертации (часть автореферата) на тему «Управление ближним электромагнитным полем в резонансных наноструктурах золото-кремний»

Реферат Общая характеристика работы

Введение и мотивация. Серия предварительных экспериментов в микроволновом диапазоне частот, предложенных в 1928 году, логически привела к развитию технологий ближнепольной микроскопии в оптическом диапазоне частот в середине 1980-х годов [1,2]. Изначально, внимание большинства исследователей было сфокусировано на оптической микроскопии высокого разрешения, что объясняется возможностью преодоления дифракционного предела при помощи этого метода [3]. Идея микроскопии ближнего поля была впервые предложена Эдвардом Хатчинсоном Сингом в его письме Эйнштейну в 1928 году [4]. Однако в последние несколько десятилетий оптические исследования ближнего поля начинают концентрироваться на квазистатических электромагнитных взаимодействиях света и вещества на субволновых расстояниях между наночастицами, что привело к развитию новых направления оптических методов исследований, таких как ближне-польное [5-7], и поверхностно-усиленное комбинационное рассеяние света [8], безызлучательный перенос энергии [9,10] и др.

Оптические исследования ближнего поля связаны с возбуждением нераспространяющегося поля во время взаимодействия света с веществом в субволновом масштабе. Это хорошо проявляется в плазмонных наноструктурах, способных сильно локализовывать электрическое поле вблизи к наноструктуре. Фактически, это изначально связано с рассеивающими свойствами наночастицы субволновой длины, помещенной в колеблющееся электромагнитное поле. Переменное электрическое поле заставляет свободные электроны когерентно колебаться. Для металлических наночастиц проводимость относительно высока, что приводит к появлению восстанавливающей силы, действующей на электроны проводимости на поверхности нано-частицы. Таким образом, образуется резонанс, вызывающий усиление поля внутри и рядом с наночастицей, определяемый как локализованный поверхностный плазмонный резонанс (ЛППР). Для одиночной плазмонной нано-частицы (Л§, Ли и т.д.) ЛППР зависит от ее формы и размера [11, 12].

Для пар индентичных наночастиц, находящихся на определенном расстоянии, локализованные поверхностные плазмоны отдельных наночастиц взаимодействуют друг с другом, что создает связывающие и несвязывающие плазмонные моды димера [13,14]. Эти две моды демонстрируют частотные сдвиги ЛППР по сравнению с модой одиночной наночастицей, который экспоненциально спадает с увеличением расстояния между ними. Таким образом, для достижения ограничения электрического поля, индуцированного ЛППР, необходимо учитывать форму, размер и расстояние между плазмон-ных наноструктур.

С другой стороны, материалы с высоким показателем преломления (Б1, Се, Бе и т. Д.) Демонстрируют электрические и магнитные Ми-резонансы, обеспечивающие усиление электрического поля внутри и рядом с наноструктурой с низкими потерями [15]. Гибридные наноструктуры, представляющие интеграцию плазмонных и диэлектрических компонентов в одну резонансную систему, могут сочетать преимущества плазмонных и диэлектрических материалов и демонстрировать множество интересных эффектов связи, таких как Фано-резонансы [16], эффект Керкера [17,18], повышенное поглощение [19] и нелинейный оптический отклик [20]. Однако все эти эффекты основаны на управлении свойствами дальнего поля. При этом эффекты усиления ближнего поля также интересны для гибридных наноструктур. Большинство гибридных наноструктур, таких как димеры и наноструктуры типа ядро-оболочка, получают путем размещения объектом статически заряженного кантилевера [20], химического синтеза [18] и измельчения ионами Не+ [21]. В этом случае довольно сложно точно контролировать ближнее поле. В результате точный контроль распределения ближнего поля остается проблемой в гибридных наноструктурах.

В данной диссертационной работе предложены и изучены новые подходы по управлению ближним полем в резонансных наночастицах металл-диэлектрик на основе золота и кремния. Кроме того, ЛППР чувствителен к форме, геометрии, расстоянию между плазмонными наноструктурами. Для этой цели предложено и исследовано несколько конструкций для управления ближним полем: (1) сферическая конусная наноантенна золото-

кремний с перестраиваемым зазором между компонентами; (2) диск-конус золото-кремний с изменением формы золотых компонентов; (3) упорядоченные массивы наноантенн с вращательной симметрией наноантенн золото-кремний с пространственным расположением. Гептамер, состоящий из 6 на-ноантенн золото-кремний, расположенных вокруг одной центральной гибридной наноантенны меньшего диаметра нижнего основания кремниевого наноконуса с расстоянием 30 нм между краями центральной и периферийной наноантенн, выбран из-за высокой симметрии вращения.

Цель данной диссертационной работы: разработка новых методов управления ближним полем в резонансных метал-диэлектрических нано-частицах.

Для достижения поставленной цели были решены следующие задачи:

Научные задачи:

1. Исследовать влияние расстояния между металлической и диэлектрической компонентами золото-кремниевой наноантенны на распределение электромагнитного поля в ближней зоне.

2. Исследовать эффект генерации второй гармоники в золото-кремниевой наноантенне в которой металлическая и диэлектрическая компоненты находятся на расстоянии друг от друга.

3. Проанализировать зависимость распределения электромагнитного поля в ближней зоне в золото-кремниевой наноантенне и ее оптических свойств в дальней зоне от формы металлической компоненты.

4. Изучить процесс управления электромагнитным полем в ближней зоне в группах золото-кремниевых наноантенн путем модификации формы их металлических компонентов.

Научная новизна данной диссертационной работы описана (но не ограничена) следующими пунктами:

1. Показано, что в золото-кремниевой наноантенне, состоящей из золотой наносферы, расположенной над верхним основанием усеченного кремниевого наноконуса, наблюдается усиление электрического поля между золотой и кремниевой наночастицами, что позволяет достигать высоких значений фактора Парселла для точечного дипольного источника в алмазной наносфере диаметром 10 нм, помещенной между золотой и кремниевой наночастицами.

2. Обнаружен эффект усиления интенсивности сигнала генерации второй гармоники в золото-кремниевой наноантенне, состоящей из золотой наносферы, расположенной через адгезионный подслой хрома на кременивом наноконусе, по сравнению с кремниевым наноконусом без золотой наносферы.

3. Установлено, что в металлодиэлектрической наноантенне, состоящей из золотого нанодиска, расположенном на верхнем основании усеченного кремниевого наноконуса, при изменении формы золотой нано-частицы от нанодиска к наносфере наблюдается изменение соотношения вкладов золотой и кремниевой наночастиц в спектр рассеяния золото-кремниевой наноантенны вплоть до устранения вклада золотой наночастицы, а также сдвиг в коротковолновую область полосы максимального усиления электромагнитного поля в ближней зоне на-ноантенны.

4. Впервые экспериментально продемонстрировано, что в гептамерах золото-кремниевых наноантенн, выполненных в геометрии золотой нанодиск-усеченный кремниевый наноконус, необратимое изменение формы золотых компонент наноантенн в гептамере под действием фемтосекундного лазерного излучения приводит к изменению пространственного положения локальных максимумов ближнего поля и усилению электромагнитного поля в ближней зоне гептамера.

Положения, выносимые на защиту:

1. В металлодиэлектрической наноантенне, образованной золотой нано-сферой, размещенной над верхним основанием усеченного кремниево-

го наноконуса, расчетные значения фактора Парселла достигают 103 для горизонтальной и 6.3 х 103 для вертикальной ориентации точечного дипольного источника, расположенного в алмазной наносфере диаметром 10 нм., помещенной в зазоре между золотой и кремниевой компонентами наноантенны.

2. Металлодиэлектрическая наноантенна, образующаяся в результате модификации фемтосекундными лазерными импульсами усеченного кремниевого наноконуса, вершина которого покрыта тонкой пленкой из золота с адгезионным подслоем хрома, может демонстрировать усиление сигнала второй гармоники более чем на два порядка по сравнению с кремниевым наноконусом без золотой наночастицы..

3. В геометрии возбуждения светом кремниево-золотой наноантенны вдоль ее оси вращения, вклад в рассеяние золотой наночастицы по отношению ко вкладу от основания наноантенны - усеченного кремниевого наноконуса - убывает при изменении формы золотой наночастицы с цилиндрической на сферическую. При этом происходит сдвиг полосы максимального усиления ближнего поля в коротковолновую область оптического спектра.

4. Необратимый сдвиг максимума в длинноволновую область в спектре рассеяния гептамера, состоящего из шести металлодиэлектрических наноантенн из усеченных кремниевых наноконусов с расположенными на их вершине нанодисками из слоев золота и хрома, размещенных в зоне ближнего электромагнитоного поля центральной металло-диэлектрической наноантенны, происходящий под действием фемто-секундного лазерного излучения, связан со спектральным смещением магнитного дипольного резонанса системы вследствие модификации формы золотых компонент наноантенн и сопровождается усилением ближнего поля, измеряемого волоконным зондом апертурного типа.

Научная и практическая значимость диссертационной работы заключается в том, что предложены и экспериментально проверены новые подходы к настройке распределения ближнего поля в гибридных резонансных

золото-кремниевых наночастицах и их массивах. Также показана возможность усиления электрического поля, сигнала второй гармоники, а также реализации высокого значения фактора Парселла в таких системах. Разработанные методы реконфигурации и настройки ближнего поля, а также гибридные наноантенны золото-кремний могут быть использованы для создания новых многофункциональных нанофотонных устройств и применяться для широкого спектра задач: от зондирования и катализа до квантовых чипов.

Надежность и достоверность полученных результатов определяется современными экспериментальными методами исследования, которые широко используются, результаты измерений продемонстрировали высокую воспроизводимость. Также показано хорошее соответствие между экспериментальными данными и результатами численного моделирования.

Реализация полученных результатов. Результаты исследований используются в качестве лекционных материалов в курсе «Single photon emission and counting: from basics to nanophotonic applications» международной магистерской программы «Нанофотоника и метаматериалы». Экспериментальные методы, описанные в диссертационной работе, используются в качестве учебных пособий в оптической лаборатории Физического факультета университета ИТМО.

Апробация. Основные результаты работы были представлены на таких международных конференциях как: 5-ая Международная Школа и Конференция "Saint Petersburg OPEN 2018 Санкт-Петербург, Россия (постер); 12-ый Международный Конгресс по Искусственным Материалам для Новых Волновых Явлений - Metamaterials 2018, Эспоо, Финляндия (постер); Международная Конференция по Нанофотонике и Метаматери-алам METANANO 2018, Сочи, Россия (постер); Международная Конференция по Нанофотонике и Микро/Нанооптике 2018, Рим, Италия (постер); IV Международная Конференция по Нанофотонике и Метаматериа-

лам METANANO 2019, Санкт-Петербург, Россия (постер); Международная школа Сигмана по лазерам 2019, Нью Йорк, США (постер, грант OSA); Quantum Science, Nanotechnology to Highlight SPIE Optics + Photonics 2019, Сан Диего, США (устный доклад, грант SPIE); 1S-brn Международный Конгресс по Искусственным Материалам для Новых Волновых Явлений -Metamaterials 2019, Рим, Италия (устный доклад); V Международная Конференция по Нанофотонике и Метаматериалам METANANO 2020, онлайн (устный доклад); 1б-ая Международная конференция IEEE по Нано/микро-инженерным и молекулярным системам (IEEE-NEMS 2021), Сямынь, Китай (приглашенный доклад), а также научные семинары в Университете ИТМО и Институте нанотехнологий CNR (Лечче, Италия).

Публикации. Основные научные результаты диссертации изложены в 8 статьях, которые индексируются в научных базах данных Scopus и Web of Science.

Личный вклад автора. Диссертация представляет собой результаты исследований, проведенных лично автором или в соавторстве. Вклад автора заключается в проведении численного моделирования и экспериментальных измерений, обработке данных, подготовке научных рукописей исследовательской работы. Все результаты, описанные в научной новизне диссертации и положениях, получены лично или при окончательном участии автора.

Объем и структура работы. Диссертация состоит из введения, пяти глав, заключения и списка использованной литературы. Общий объем диссертации 89 страниц, в том числе библиография - 104 ссылки. Работа содержит 40 рисунка и S таблицы, размещенные в главах.

ОСНОВНОЕ СОДЕРЖАНИЕ РАБОТЫ

В первой главе описана актуальность проведенного исследования и представлен обзор литературы, теоретический анализ, эксперименталь-

ные подходы, и продемонстрированы актуальные проблемы и мотивация диссертационной работы.

Во второй главе представлены описания методов экспериментальных исследований и численных расчетов, относящиеся к диссертационной работе.

В третьей главе диссертации обсуждается влияние расстояния между металлической и диэлектрической компонентами золото-кремниевой наноантенны на распределение ближнего поля. Внутри небольшого зазора создается сильное локализованное электрическое поле, которое применяется для усиления излучения источников одиночных фотонов и сигнала ГВГ.

Гибридная наноантенна с определенным зазором схематически представлена на вставке к рис. 1. Усеченный кремниевый конус имеет диаметр при основании и высоту 190 нм. Верхний диаметр конуса составляет половину диаметра нижнего. Такая конструкция позволяет достичь магнитного дипольного резонанса и электрического дипольного резонанса в видимом диапазоне длин волн. Диаметр золотой сферы равен верхнему диаметру конуса (95 нм) для стабильности системы. Зазор между кремниевым конусом и золотой сферой варьируется в диапазоне 10-45 нм.

Рисунок 1 — (я) Усиление поля в зависимости от ширины зазора между кремниевым нанодиском и золотой наночастицей на длине волны 637 нм. (Ь) Распределение электричексого Е и магнитного Н полей на длине волны 637 нм. Проекция YZ

Для расчета коэффициента усиления электрического поля по отно-

шению к зазору проведено численное моделирование на длине волны бес-фононной линии отрицательно заряженного центра окраски азот-ваканси (КУ-) в наноалмазах (637 нм.). Электромагнитная волна падает на гибридную наноантенну под углом 68 градусов относительно оси симметрии. На рис. 1 показаны распределения электрического поля и магнитного поля на длине волны 637 нм. Как можно видеть из рисунка, вектор электрического поля осциллирует в плоскости падения, что приводит к эффективному возбуждению резонанса при ТМ-поляризации, и, таким образом, электрическая энергия сильно локализуется в зазоре. Такая сильная локализация поля может применяться для связи с различными квантовыми излучателями для усиления излучения, которое характеризуется фактором Парселла. В качестве примера был взят алмаз с дефектными центрами окраски КУ-(п= 2,4), занимающий зазор гибридной наноантенны, и для вычисления фактор Парселла для горизонтальной и вертикальной ориентации диполя (длина диполя 5 нм, см. рис. 2) была применена формула согласно [22]:

_ 1т[Сгг(го,го,ш)} _ Де[^та] (1)

_ 1т[С°1 (го, го,и)} _ Яе( ) где Re[Z¿n] и Re[Z0,¿n] показывают действительную часть входного импеданса электрического диполя, связанного с гибридной наноантенной, и в свободном пространстве, соответственно.

Как и ожидалось, комбинация резонансных диэлектриков и плаз-монных наночастиц приводит к сильному усилению электрического поля и, таким образом, к увеличению фактора Парселла. Моделирование демонстрирует, что гибридная наноантенна демонстрирует резкий резонанс фактора Парселла на длине волны 637 нм для обеих ориентаций диполей и демонстрирует значение 103 и 103'8 для горизонтальной и вертикальной ориентации, соответственно.

Основной вклад в усиление сигнала фотолюминесценции вносит только радиационный фактор Парселла. При вычислении радиационного фактора Парселла по формуле Раюегаде _ 2/3^ог + 1/3^ег было получено значение 300, которое вполне конкурентоспособно для диэлектрических или плазмонных наносистем [18,23-25].

400 500 600 400 500 600

wavelength[nm] wavelength [nm]

Рисунок 2 — Коэффициент Парселла наноантенны, состоящей из золотой сферической

наночастицы, наноалмаза и кремниевого диска для случаев (a) горизонтально и (b) вертикально ориентированных дипольных моментов наноалмаза

Чтобы экспериментально исследовать влияние увеличения электрического поля в зазоре, мы изготавливаем систему из численного моделирования с помощью следующего подхода к изготовлению. Коммерческий титан-сапфировый лазер (центральная длина волны 800 нм, rpuise ~ 100 фс, частота повторения 80 МГц) используется для изготовления этого типа гибридных наноантенн с определенным зазором Au-Cr-Si гибридной нано-антенны, изготовленной методом лазерной литографии, как это показано на рис. 3 a-c. Тонкий слой Cr (2 нм) служит для прочного сцепления кремния и золота. При применении лазера с интенсивностью 845 ГВт/см2 с частотой следования импульсов 50 Гц золотой диск достигает температуры плавления, и форма меняется на сферу. В соответствии с исследованиями процесса модификации непрерывной волной [26], Cr растягивается и покрывает частицу золота, чтобы сохранить объем золота неизменным, и формирует небольшой зазор между золотом и кремнием (см. рис. 3b, так называемая щелевая наноантенна). Это небольшое отделение золотой частицы от кремниевого конуса можно объяснить преобразованием поверхностной энергии в кинетическую энергию и, таким образом, подъемом центра масс в золотой частице, что хорошо изучено в [27] с треугольными наночастицами золота на поверхностях графита и стеклянной подложки. Когда интенсивность лазера увеличивается до 1200 ГВт/см2, золотая сфера получает достаточно

энергии, чтобы оторваться от кремниевого конуса, предположительно захватив некоторую часть Сг, оставив на поверхности стеклянной подложки усеченный кремниевый конус с небольшими остатками хрома на верхнем основании конуса, названную R-конусом (см. рис. 3с).

(d) 1.0 Z5

с/5

ш 0.5

с си

В го

о:

0.0

400 450 500 550 600 Raman shift (cm1)

Рисунок 3 — Изображения сканирующей электронной микроскопии (СЭМ) наноантенн, облучаемых лазером. Слева направо интенсивность накачки увеличивается: (а) в отсутствии облучения, (b) образование зазора между сферической наночатицей и диском, (c) сферичекая наночастица отделилась от нанодиска под дейстивем накачки. Длина масштабного отрезка 200 нм. (d) Спектры Рамановского рассеяния до и после

облучения наноантенн

Поскольку кремний имеет гораздо более высокую температуру плавления, чем золото, форма кремниевого конуса не изменяется. Тем не менее, кремниевый конус является аморфным до процесса изменения формы и демонстрирует широкополосный пик в спектрах КР с центральным значением 480 см-1. После модификации фемтосекундным лазерным излуче-

нием щелевая наноантенна и Я-конус демонстрируют гораздо более узкий пик в спектрах КР при 521.5 см-1, как показано на рис. Данный пик соответствует случаю колебания ТО моды решетки поликристаллического кремния. Как известно, наноструктуры кристаллического кремния обеспечивают нарушение локальной инверсной симметрии на границе раздела и высокую вероятность достижения эффективной ГВГ. Для оценки влияния усиления электрического поля на ГВГ были найдены условия облучения, которые полностью устраняют золотую составляющую из наноантенны и не разрушают кремниевую (1200 ГВт/см2). Такие наночастицы демонстрируют гораздо более узкий кристаллический пик при 521.5 см-1, поэтому их также можно использовать для наблюдения эффекта ГВГ.

На рис. 4а показаны спектры ГВГ, полученные от щелевой наноан-тенны, Я-конуса и пленки кремния толщиной 100 нм с максимально достижимой интенсивностью. Максимальная интенсивность в каждом случае достигается вблизи порога повреждения, который определяется изменением в спектре линейного рассеяния. Максимальная интенсивность сигнала ГВГ щелевой наноантенны на 2 и 3 порядка выше, чем у Я-конуса и пленки кремния, соответственно. Чтобы объяснить разницу между порогом повреждения, выполняются численные расчеты для проверки тепловых свойств этих двух структур. Анализ как непрерывного (С^ лазера, так и импульсного лазера доказывает, что индуцированный лазером нагрев более выражен в Я-конусе, чем в щелевой наноантенне.

Кроме того, на рис. 4Ь показана зависимость интенсивности ГВГ от плотности мощности лазерного излучения для щелевой наноантенны и Я-конуса в логарифмическом масштабе, демонстрируя почти квадратичный наклон для Я-конуса и неквадратичную зависимость для щелевой наноан-тенны за счет статического поля на внешней стенке диэлектрического конуса (эффект ЭИГВГ).

Когда на материал воздействует статическое электрическое поле, восприимчивость второго порядка х(2) модифицируется за счет статического электрического поля через восприимчивостью третьего порядка х(3) которая может быть выражена следующим образом [28]:

Intensity (GW/cm )

Рисунок 4 — (a) Интенсивность генерации второй гармоники на наноантенне, R-конусе и кремниевой пленке от длины волны накачки. (b) Зависимость интенсивности генерации второй гармоники от мощности накачки

I (2ш) = |Х(2) + Х(3)^С|2/2М (2)

Таким образом, зависимость интенсивности ГВГ от интенсивности возбуждающего излучения больше не является квадратичной, и это называется эффектом электрически-индуцированной ГВГ (ЭИГВГ). В нашем случае горячие электроны генерируются золотой сферой посредством многофотонного поглощения и переходят на кремниевый конус, вызывая разделение зарядов на границе кремниевого усеченного конуса с воздухом, которая в свою очередь покрыта естественным слоем оксида (8ь8Ю2). Образовавшееся статическое электрическое поле возбуждает эффект ЭИГВГ в щелевой наноантенне с неквадратичной зависимостью (степень полиномиальной функции ~ 4). Таким образом, усиление электрического поля в зазоре, а также статического электрического поля модифицируют ГВГ в таких системах, что делает ее очень перспективной для нанофотонных элементов преобразователей частоты.

В четвертой главе описывается роль формы золотого компонента и пространственного расположения наноантенн в управлении ближним по-

лем в различных количествах гибридных наноантенн. Локализованный поверхностный плазмонный резонанс (ЛППР) можно точно контролировать с помощью воздействия фемтосекундным лазерным излучением и увеличивать ближнее поле для гибридной наночастицы на определенной длине волны.

Сначала проводится исследование эффекта настройки путем расчета спектров рассеяния и распределения электрического поля для одиночной гибридной наноантенны на разных этапах модификации (см. рис. 5). Размер кремниевого конуса остается неизменным, а диаметр золотого диска равен нижнему диаметру конуса. Плоская монохроматичная электромагнитная волна излучается сверху вдоль оси симметрии наночастицы.

Wave length i

Рисунок 5 — Изменение ближнего поля - (а) Спектры рассеяния кремниево-золоых наноантенн (красная линия) и отдельно плазмонных наночастиц соответсвующей формы (синия линия). (b) Распределение электричексого поля E на резонансных длина волн: (1) 910 нм, (2) 680 нм, (3) 750 нм, (4) 660 нм, (5) 680 нм и (6) 590 нм

Для исходной гибридной наночастицы с геометрией диска металлической части положение ЛППР резонанса находится на длине волны 910 нм (резонанс 1 на рис. 5) и соответствует резонансу золотого диска, демонстрируя коэффициент усиления электрического поля до 12. Еще два резонанса на длинах волн 690 нм и 580 нм соответствуют магнитному дипольному (МД) и электрическому дипольному (ЭД) резонансу кремниевого конуса. Поскольку ЛППР весьма чувствителен к форме и размеру плазмонной на-

ноструктуры, когда золотой диск превращается в золотую чашку, ЛППР золотого компонента претерпевает сдвиг резонанса в синюю область с 910 нм к 750 нм (см. резонанс 3 при 750 нм). Распределение электрического поля в резонансе 3 на рис. 5 показывает коэффициент усиления электрического поля до 9.5. Кроме того, когда золотая чашка превращается в золотую сферу, ЛППР золотого компонента смещается до 590 нм, почти перекрываясь с ЭД резонансом кремниевого конуса, и интенсивность сильно уменьшается. Распределение электрического поля в резонансе 6 подтверждает это и показывает сильную локализацию поля в кремниечвом конусе и рядом с золотой сферой.

о.о [—.---.-.-.---.-.-,---.-:

500 600 700 800 900 1000

\А/ауе1епд1:Ь (пт)

Рисунок 6 — Фактор Парселла кремниево-золотой наноантенны для разных форм золотой наночатицы. Ширина зазора между золотой наночастицей и кремниевым диском 145 нм (красная линия) и 125 нм (зеленая линия)

Поскольку локализация в ближней зоне сильно влияет на фактор Парселла, мы переходим к изучению влияния изменения формы на фактор Парселла для золото-кремниевой наноантенны. На рис. 6 показаны результаты моделирования фактора Парселла для различных конфигураций гибридной наноантенны. Как можно видеть, более близкое расстояние между

Список литературы диссертационного исследования кандидат наук Сунь Яли, 2021 год

p • r

4^eoemr3

(1.3)

p = 4^ eg ema3 e+ 2m £q (1.4)

6 + 2 £ m

In this case, we can clearly observe the induced dipole moment within the sphere (Eq. 1.3) and its intensity is proportional to the intensity of the applied field (Eq. 1.4). As is well-known, the dipole moment can be utilized to define the polarizability a though p = eQemaEq, therefore we can describe the polarizability a as [33,34]:

a = ^ (1.5)

6 + 2 Cm

when the |e + 2e m| is at the minimum value, the polarizability is resonantly enhanced. As a result, for a Drude-model described plasmonic particles (Eq. 1.6), for small values of the imaginary part of e (slowly-varying ones also count) adjacent to the resonance, we can simplify the real part of e as a function of the frequency u as [33,34]:

u2

e(u) = 1 - 2 p (1.6)

v y u2 + iqu '

Re [e(u)] = -2 em (1.7)

This phenomenon is named as Frohlich condition and the corresponding resonance in the oscillated field is called dipole surface plasmon resonance of the plasmonic nanoparticle. For a Drude-model described plasmonic sphere placed in air (em =1), the Frohlich condition is satisfied at uQ = up/\/3. This identifies that plasmonic resonance is strongly related to the dielectric medium which can be applied for sensing.

Further we turn to calculate the electric field distribution via E = —V 0 and obtain the fields inside and outside the sphere as [33]:

Ein = —2— Eq (1.8)

6 + 2 £ m

P = p 3n(n ■ p) - p 1

Emd = Eq + 4^ r3 (1.9)

As predicted, the resonance in a also results in the enhancement of the internal and dipole fields. Thus the proper system design lies in strong electric field enhancement achieved at the plasmonic resonances.

So far, we have evaluated the case when the nanoparticle is located in a static electric field. For an excitation of plane wave as E(r,t) = E0e—%ut, the dipole moment is oscillated with the fields as p(t) = e0emaE0e—lLVt. The dipole radiation results in scattering the illuminated plane wave by the nanoparticle which can be identified as a point dipole radiation.

For an electric dipole, the total magnetic field H(t) = He—'Mt and electric field E(t) = Ee—%ut are expressed as [33,34]:

nh2 pikr 1

H = ^ (n X p) — (1 - —) (1.10)

4^ ' r ikr v y

1 ^ikr 1 ^^

__{fc> x p x n_ + [3„(„ • p) - P)](-J _ _

ikr"

in which k = 2n/\ and n is the unit vector in the direction of studied point P. For regions kr ^ 1 (near-field), the electric field is similar to the static electric field as

3n(n • p) — p 1

4^eoem r

The corresponding magnetic field is expressed as:

e = 3"7_ •") — ^ (1.12)

H = £(n x P) 1 <1-13)

It should be noted that in the near-field zone, the electric field dominates the majority due to the much lower intensity of magnetic field which is identified as sje0/[i0(kr). Especially, when kr ^ 0 (static field), the magnetic field disappears.

On the other side, when kr ^ 1 (radiation region), the electric and magnetic field can be described by spherical-wave form as [33]:

nh2 pikr

H = ^ (n x p) — (1.14)

4^ r

E = A [^H x n (1.15)

V e o £ m

Apart from the electric and magnetic field distribution, scattering and absorption induced by the nanoparticles are also important for design optical systems and they can be calculated by Poynting-vector and expressed as [33, 35]:

^=6- M2 = i2 (1.16)

6^ 3 e + 2 e m

Cabs = k/m[a] = 4^a3/m[ 6 ~ €m ] (1.17)

£ + 2 £ m

It should be noted that our derivations are not conducted under the assumption it is a plasmonic nanoparticle, thus Eq. 1.16 and Eq. 1.17 are suitable for dielectric nanospheres as well. As can be seen, Csca rc a6, thus it is quite difficult to select small nanoparticles out of the environment of larger scatters. However, this is possible to realize with near-field characterization methods due to diffraction limit overcoming. Different near-field characterization methods are presented in the next section.

1.2 Near-field characterization and applications

Compared with the far-field resolution, near-field characterization can overcome the diffraction limit and provide much more information. Several techniques are proposed: scanning near-field optical microscopy (SNOM) or near-field scanning optical microscopy (NSOM) [36,37], photon emission electron microscopy (PEEM) [38,39], atomic force microscopy (AFM) [40], scanning tunnelling optical microscopy (STOM) [41], photon scanning tunnelling microscopy (PSTM) [42] etc. It should be noted that simple near-field microscopy images represent the information where are the small objects such as molecules, ions.

An example of single Ca2+ transmembrane protein in liquids is shown in Fig. 1.2 a-c [43]. The confocal image is blurry due to the diffraction limit. However, by localizing the laser radiation to 50 nm via a gold nanoparticle,

the resolution of single membrane proteins can be reduced to 5-10 nm. In addition, by integrating the Near-field microscopy and Raman spectroscopy, so-called near-field Raman spectroscopy, it can provide information on what is under investigation [5,7]. One example of single chain identification in singlewall carbon nanotube by near-field mapping at the vibration wave vector of single chain is presented in Fig. 1.2d-f. Theoretical and experimental works demonstrate the Raman enhancement is mainly decided by the plasmonic nanoantenna for detection rather than the specific phonon eigenvector. This method can be widely applied to study the defects, strains and dopants with high resolution.

Figure 1.2 — Near-field imaging and coupled with spectroscopy (a-c) single Ca2+ transmembrane proteins in liquids, confocal image and near-field characterization comparison. Adopted from [43]; (d-f) Near-field Raman scattering of single chain in nanocarbon, confocal image, near-field Raman scattering image and measured Raman

spectra, adopted from [7]

Except for the imaging and integration with spectroscopy, near-field can be also applied to enhance the normalized local density of states characteristic as Purcell factor in the case of nanoantenna coupled with quantum

emitters [44] and sensing [45,46], as shown in Fig. 1.3. A single-photon source is placed in the highly localized electric field generated by radiative nanoanten-nas, which provides a near-field enhancement for fast and pure single-photon emission with a brightness exceeding 109 photons/sec [44]. For gold bow-tie nanoantenna, we observe enhancements of the fluorescence from a single molecule placed in the gap of the bow-tie nanoantenna up to 1340 times [45]. Plasmonic nanoparticle-on-mirror geometry is also applied to integrate single molecule leading to decreasing to 90 meV for single molecules—matching quantitative models [46].

Figure 1.3 — (a) Near-field enhanced single-photon emission, adopted from [44]; (b) single molecule coupled with a plasmonic cavity, adopted from [46]

To achieve the above applications of near-field, the realization of effective near-field interaction in plasmonic nanoantennas is the key problem. For a single nanoparticle, we can utilize the equations in section 1.1 to predict the electric and magnetic fields. However, for more complicated systems, we perform numerical modelling and experimental demonstration in far-field properties to search for the LSPR and manipulate the near-fields. Thus we turn to investigate different approaches to control near-field through far-field detection in nanoscale in section 1.3.

1.3 Approaches to control near-field in nanoscale

Section 1.2 demonstrates the advantages of near-field microscopy compared with far-field images and strong near-field localization induced applications. Thus we can design optical systems to achieve near-field confinement

and apply them to photonic applications. Even section 1.1 presents the analytical method to estimate the LSPR of a nanosphere and additionally describe the electric and magnetic fields. However, spherical nanosphere is quite difficult to fabricate as desire. Therefore most applications are based on plasmonic disks, nanorods and bow tie nanoantennas. Numerical simulations are conducted to understand the LSPR of these nanoantennas and manipulate the strongly localized electric field. This section is mainly focused on how to control the near-field of plasmonic nanostructures in nanoscale. The key element is nanoantenna which can strongly localize the light in the near-field region.

« 1,4 1,5 1 8 2.0 2.2 400 450 500 550 SM sso 700 750

w Photon energy (eV) Wavelength (nm)

Figure 1.4 — Shape influence on the plasmon resonance of plasmonic nanoparticles (a-b)

SEM images and scattering spectra of gold nanorods with different sizes, adopted from [11];(c) scattering spectra of spherical, pentagonal and triangle silver nanoparticles

with SEM images [12]

We start from the exploration of shape and size influence on the LSPR which can be observed in scattering spectra measured by confocal dark-field microscopy for a single nanoparticle. Fig. 1.4 shows the scattering spectra of different sizes of gold nanorods and three shapes of silver nanoparticles (spherical, pentagonal, triangle). As can be seen, the surface plasmon resonance is strongly related to the shape and size of the plasmonic nanoparticles and the resonance shift can cover the whole visible range. For small particles, the interaction of LSPR can be simplified as the interaction of dipole interaction.

The second stage is to investigate the LSPR in pairs of identical

Inter-particle gap (nm)

Figure 1.5 — Plasmon hybridization of dimers (a) theoretical illustration of bonding and anti-bonding plasmon shift, adopted from [14]; (b) experimental demonstration of plasmon

shift in the gold disk, adopted from [13]

nanoparticle with a certain distance (d) [13,14]. The localized surface plasmon of individual nanoparticle interacts with each other and construct the bonding and anti-bonding dimer plasmon. These two modes exhibit an LSPR shift compared to the single nanoparticle, so-called plasmon shift with a given angular momentum (see Fig. 1.5). The dimer plasmon is strongly dependent on the distance separating the two nanoparticles. As the distance increases, the plasmon shift exponentially decays. Moreover, for a small single particle, we can treat the dimer as a dipole interaction. According to Eq. 1.5, the polariz-ability a rc a3 and the decay of the plasmon near-field rc 1/r3. Thus a dipole-coupled model can be qualitatively solved by dipole-dipole approximation. Researchers qualitatively demonstrate the decay length is approximately 0.2 of the nanoparticle size unit with different shapes, sizes and the surrounding medium [13]. A more complicated system such as one-dimensional chain [33] and hollow structures [30] are also popular designs to achieve plasmon hybridization. Since these configurations are not under consideration in this thesis work, detailed information is neglected here.

On the other hand, distance control is also important for chemical reaction induced hot spots of bow-tie nanoantenna by He+ -ion milling lithography [21,49]. Besides, the distance between the individual particle plays an important role in hot-spots switching [47]. Distance control in oligomers be-

Figure 1.6 — (a) He+ -ion milling lithography induced hot spots in bow-tie nanoantenna, adopted from [21]; (b) hot spots switching in gold trimer, adopted from [47]; (c) scattering spectra of heptamer with different center-to-center distance, adopted from [48]

tween the central particle and the peripheral particles is also studied to achieve collective effects and high local electric field confinement [48].

Upon the nanoantenna design, geometry and distance take the majority part in near-field control. However, the designed system is quite sensitive to the polarization of the excitation. Thus we reveal the polarization dependence in the near-field configuration in Fig. 1.7. Oligomers are of special interest in this thesis work. Among them, heptamer is the most popular configuration due to its symmetry properties. As can be seen, radial and azimuthal polarization result in bonding and anti-bonding of surface plasmon modes. For trimer, different orientation of linear polarization, left-hand circular polarization (LCP), right-hand circular polarization (RCP) are applied to map the near-field configuration. In addition, strong electric field confinement is suitable for sensing. Hereby integrating the near-field manipulation with chirality, the optical system can be utilized for polarization-sensitive photochemistry and biology detection.

Figure 1.7 — (a) Bonding and (b) anti-bonding modes in heptamer under azimuthal and radially polarization, adopted from [29]; (c) near-field under LCP and RCP of the nanorod, adopted from [37]; (c) near-field of trimer under linear and circular polarization, adopted

from [40]

So far we review the role of shape, distance and polarization in near-field control of plasmonic nanostructures. However, an alternative way to achieve strong near-field localization relies on another type of material-all dielectrics. All-dielectric materials are quite interesting and important in nanophotonics due to their electric and magnetic Mie-resonances in the visible range with low losses [15]. Thus they can confine the electromagnetic wave within and around the nanostructures. Besides, unidirectional scattering can be realized in a single dielectric nanoantenna as the interference of magnetic and electric dipole responses oscillated simultaneously in the nanostructure with comparable amplitude [50]. In contrast to plasmonic nanostructure which enhances the nonlinear effects by near-field enhancement, the resonances of all-dielectric nanostructures induce a mode volume that can provide interfaces leading to higher conversion efficiencies of nonlinear effects [15].

In this case, we believe hybrid metal/dielectric nanostructures can com-

bine the advantages from plasmonic components and dielectric counterparts simultaneously and achieve more efficient and pronounced optical effects compared to pure plasmonic or dielectric nanosystems. Therefore we further represent the potential optical effects achieved in hybrid metal/dielectric nanos-tructures in the next section.

1.4 Hybrid metal/dielectric systems

High metal/dielectric nanostructures have attracted intense intention in the last decade due to the combination of the benefits of metal and dielectric parts. Thus they can manipulate the light-matter interactions via the strengths of each part and the coupling between them.

Integration of plasmonic and dielectric nanostructures typically lies in core-shell, dimer and substrate-induced mirror configuration. These configurations can be fabricated by pick and place (mostly for dimers), chemical synthesis, lithography etc. There are two principles in hybrid metal/dielectric systems designs: (1) by carefully tuning the geometry and arrangement, overlapping the resonances from dielectrics and plasmonics components lead to plenty of broadband effects; (2) placing the dielectric components in a plas-monic environment and utilizing near-field enhanced dielectric effects.

Overlapping effects include unidirectional scattering [17,18], Fano resonance [16], enhanced absorption [19] and nonlinear effects [20], etc., as demonstrated in Fig. 1.8a-e. For a dielectric nanoparticle, unidirectional scattering is limited to a single wavelength which is the electric and magnetic dipole/quadruple crossing point in multipole expansion spectra. However, for hybrid metal-dielectric nanostructures, we can obtain broadband unidirectional scattering by overlapping the electric dipole resonance from the plas-monic part and magnetic dipole resonance in the dielectric part (see Fig. 1.8a-b). This overlapping principle also suits the broadband absorption and nonlinear enhancement in Fig. 1.8d-e. Besides, Fano resonance, originating from the interference of a broad and a narrow mode, is much easier to be achieved in a single hybrid nanoparticle due to the overlapping of broadband dipole resonance from plasmonic nanostructure and quadruple modes from dielectrics

Figure 1.8 — Enhanced effects by resonance overlapping from metal and dielectric components: (a) theoretical unidirectional scattering, adopted from [17]; (b)experimental

unidirectional scattering, adopted from [18]; (c) Fano effect in core-shell nanostructure [16];(d) broadband absorption in core-shell nanoparticle, adopted from [19]; (e) broadband SHG enhancement, adopted from [20]. Nonlinear enhancement by placing dielectrics in hot spots of plasmonics oligomers (f), adopted from [51]

(see Fig. 1.8c).

Furthermore, Even without resonance overlapping between the dielectric and plasmonic counterparts, the plasmonic component can boost the anapole modes in dielectric nanostructure with induced nonlinear enhancement [52,53]. Moreover, Fig. 1.8f shows an alternative method to enhance the nonlinear signal by placing the dielectric nanowire in the hot spots of gold oligomer [51]. As can be seen from the figure, the SHG intensity from the hybrid part is three orders higher than that from the pure dielectric wire. This motivates us to study the electric field enhanced nonlinear effects in this thesis work.

However, all the effects mentioned above mainly concern the far-field properties of hybrid metal/dielectric nanostructures. Near-field enhanced effects are also interesting for hybrid nanostructures. Most hybrid nanostruc-

tures such as dimers and core-shell nanostructure are obtained through pick and place, chemical synthesis and He+ -ion milling [18,20,21]. It is quite difficult to precisely control near-field in this case. As a result, precisely near-field control and tuning remain a challenge in hybrid nanostructures. Thus we dedicate to study the near-field confinement in hybrid nanostructures taking shape, distance and polarization, resonant condition into account for optimal design.

1.5 Chapter conclusions

This chapter mainly describes the physics of LSPR for a single plas-monic nanoparticle and further investigate the plasmon hybridization of dimer and oligomers via shape, distance and polarization control. Besides, impressed by plenty of broadband effects in hybrid metal/dielectric nanostructures, it is figured out the near-field control in hybrid nanoantenna remains a challenge. This motivates us to explore the optical near-field control in hybrid nanos-tructures numerically and experimentally.

CHAPTER 2. EXPERIMENTAL TECHNIQUES AND NUMERICAL

METHODS

This chapter is dedicated to the fundamental research methods including main fabrication methods, experimental characterizations and numerical simulations applied to this research. Section 2.1 represents the fabrication methods includes lithography and femtosecond laser treatment (section). Section 2.2 shows the experimental characterization methods includes dark-field scattering, Raman scattering, scanning near-field optical microscopy and second harmonic generation. Section 2.3 discusses the typical numerical simulation solvers and comparison for CST Microwave Studio.

2.1 Fabrication methods

Samples are the basis for experimental studies as the bricks for the buildings. Typical fabrication methods for single-particle resolution include lithography, chemical methods, dewetting, laser-assisted methods, green printing, focused ion/electron beam and so on. For nanophotonics applications, resonant properties of the studied structures are essential, and it can be tuned in the visible range by controlling the size and shape during fabrication. Thus dimension control, repeatability, resolution, position control, fabrication complexity, productivity are quite important to evaluate the fabrication methods. These properties for high refractive index materials fabrication by typical methods were summarized in [54]. In this dissertation work, samples are fabricated in two ways: lithography and femtosecond (fs-) laser treatment. These important issues above are taken into account and the fabrication methods are introduced sequentially.

2.1.1 Lithography fabrication

Lithography is the most reliable and controllable method for fabricating single particle as well as large array metasurface. By a combination of

lithographic processes, it can precisely realize the nanostructure of different materials with multi-shapes, such as disk, truncated cones, hollow etc. with high repeatability, and the minimum size can be achieved as small as 10 nm.

Figure 2.1 — Fabrication process of the gold-silicon disk-cone nanoantenna by lithography. (a) e-beam lithography; (b) Metal films deposition; (c) lift-off; (d) chemical etching

Fig. 2.1 demonstrates the fabrication process of the most studied goldsilicon disk-cone nanoantenna through lithography methods. As can be seen from Fig. 2.1a, 200 nm thickness of a-Si: H film is deposited on a glass substrate by plasma-enhanced chemical vapor deposition of SiH3 gas. Then with the help of PMMA mask, disk array shapes are produced by electron beam lithography with a controllable distance between each other. This procedure follows Cr 1 nm) and Au 20 nm) layer deposition (Fig. 2.1b). Except for the metal films in the disk array, other parts of plasmonics materials are removed by lift-off procedure, forming the Cr/Au disk on silicon film (Fig. 2.1c). Finally, the silicon layer turns into truncated silicon cone by chemical etching (SF6 and O2 gases, see Fig. 2.1). It should be noted that this method has high repeatability and is able to fabricate a large array of such gold-silicon disk-cone nanoantenna simultaneously. Moreover, the distance between the gold-silicon nanoantenna can be easily tuned for different applications, for instance, for single-particle spectroscopy, the distance between the gold-silicon nanoantenna is set more than 1 ym, but for data storage, the distance between the gold-silicon nanoantenna is set around 500 nm. Except for single goldsilicon nanoantenna, pairs of gold-silicon nanoantenna, heptamer oligomers in this dissertation work are all fabricated in the same way. Although lithography is a complex technique and requires expensive equipment as well as rich experience, it is still quite competitive and reliable for nanophotonics research.

This part is carried out by Dr. Ivan Mukhin at Saint Petersburg Academic University.

2.1.2 Femtosecond laser treatment

Lithography is a powerful tool to fabricate different samples, however, one limitation of lithography is spherical nanoparticles fabrication. Spherical nanoparticles are quite interesting for fundamental studies as well as applied science. Besides, the physical properties of nanospheres can be analytically calculated (see section 1.1), which can save plenty of time for designing functional nanosystems. Thus two ways are proposed to fabricate spherical nanoparticles: first, based on the gold-silicon disk-cone nanoantenna fabricated by lithography, fs-laser reshaping is applied to fabricate semi-spherical and spherical nanostructures; second, fs-laser ablation of nanospheres from films.

Figure 2.2 — Reshaping process of the hybrid gold-silicon nanoantenna. (a) one hybrid nanoparticle (b) pairs of hybrid gold-silicon nanoantenna; (c) SEM images of reshaping

process, adopted from [26]

Fs-laser reshaping is applied to modify the morphology of the gold component without changing the shape of the dielectric cone in gold-silicon disk-cone nanoantenna. It is widely utilized to reshape single hybrid nanoparticle, a pair of hybrid nanoparticles (see Fig. 2.2a-b) and different oligomers. The principle of fs-laser shaping (see Fig. 2.2c) relies on the relatively lower

melting temperature of gold (1064 °C) compared to silicon (1414 °C). As can be seen from the scanning electron microscope (SEM) images, the gold disk starts to melt when the absorptive heat induced by fs-laser exceeds the melting temperature threshold (for single hybrid nanoparticle, laser fluence « 28 mJ/cm2), then with the increasing of laser fluences, the gold disk changes the shape to cup and finally turns into spheres. However, once the laser fluence is too high, the sphere can jump away and leave a silicon cone with a small amount of Cr (Si-Cr cone). It should be noted that the thin Cr layer can not only be used to adhere to the silicon and gold layer, but also surround the gold part during the reshaping process. In this case, the volume of the gold component remains the same and possesses the possibility to numerical study the modified structure. Another feature of the fs-laser reshaping method is repetition rate, which will be detailed described in section 3.3. Besides, It is experimentally proved that fs-shaping has good repeatability no matter with piezo stage or manually conducted. Fs-laser reshaping can be applied to modify different plasmonics materials (Ag, Al, etc.) and provides one way to tune the optical properties of hybrid nanostructures after lithography.

Figure 2.3 — Laser ablation of c-Si nanospheres from silicon films, adopted from [55]

Fs-laser reshaping suits for hybrid nanostructures, but alternative easier method-laser ablation is proposed for single sphere fabrication on different substrates. Here a 100 nm a Si:H film is deposited on a glass substrate by plasma-enhanced chemical vapor deposition of SiH3 gas for fs-laser ablation. Fs-laser is a commercial Yb-doped solid-state ultra-fast laser centered at 1050 nm, and the repetition rate is controlled by a Pockels cell-based pulse picker. A receiving substrate (clean glass, Au film, etc.) is placed around 50 ym under the silicon film. While the laser beam focuses on the silicon layer by a Mi-tutoyo Apochromatic NIR Objective with 10 times magnification (NA=0.26), the spherical silicon nanoparticles are ablated and deposited on the receiving substrate (see Fig. 2.3). By carefully controlling the laser spot diameter D/aser = 1.22A/NA and laser fluence, the diameter of the spheres can be set in the range of 50-200 nm. Laser ablation method can be applied to fabricate single spheres of different materials, hybrid sponge spheres of multi-materials. Considering its simplicity, it is widely applied in nanophotonics research.

2.2 Experimental characterization methods

Scattering is the most common light-matter interactions in which the light alters propagating direction and perhaps its frequency after interacting with the materials. That means the total number of photons of the incident light remains the same, but the number of photons propagating along the original direction decreases as light is re-directed in many other directions. If one only concentrates on forward direction propagation, then scattering plays the same role as absorption in light intensity attenuation. By checking whether the scattered light frequency changes or not, scattering can be divided into elastic scattering and inelastic scattering. For elastic scattering, the scattered light remains the same frequency, this is what described in section 2.2.1. Inelastic scattering changes the light frequency and details are summarized in section 2.2.2. Scattering is far-field properties and near-field properties are also interesting for enhanced SERS, photoluminesence, etc. Near-field characterization method is introduced in section 2.2.3. Nonlinear experimental demonstration is presented in section 2.2.4.

2.2.1 Dark-field scattering

Elastic scattering can be observed very often in daily life, such as blue sky, rainbow, beautiful Peacock feathers and so on. By taking into account the diameter of the scatter and the incident wavelength, a dimensionless size parameter /3 = ^D/A is defined. If /3 ^ 1, i.e., the diameter of the scatter is very small compared to the incident light wavelength, it's defined as Rayleigh scattering. If /3 ~ 1, i.e., the diameter of the scatter is comparable to the wavelength of the light, it's defined as Mie scattering. If /3 ^ 1, i.e., the diameter of the scatter is much larger than the wavelength of light, it's defined as geometric scattering. Nanophotonics studying in this dissertation work is mostly concentrated on the light-matter interactions in the visible range, and the diameter of the nanoparticles studied here are in the scale of several hundreds of nanometer, so Mie scattering is the main focus point.

Reflection

Halogen Lamp

Polarizer

Glass

Figure 2.4 — Schematic of dark-field scattering

For spherical nanoparticles, Mie theory is an efficient tool to analytical predicts the resonant properties of the nanospheres. By solving the Maxwell equations and substituting the size and material properties (complex refractive index), the scattering cross-section (SCS) of the nanospheres can be calculated

with multipole expansion to analyse the electric dipole (ED), magnetic dipole (MD), electric quadrupole (EQ), magnetic quadrupole (MQ) and even higher order of modes if necessary. For arbitrary shapes (cylinder, cubic, cup, etc.), it's difficult to analytically calculate the SCS, however, several commercial software including COMSOL(finite element technique), CST Microwave Studio (finite integral/element technique) and Lumerical (finite difference technique) provides a convenient and efficient way to calculate the scattering by building an identical morphology model with proper boundary conditions. Most of the projects in this dissertation work are calculated in CST Microwave Studio. Details about CST solvers and calculation characteristics are described in section 2.3.

As for the experimental demonstration of scattering, a confocal dark-field (DF) microscopy is applied to measure the scattering at an oblique incidence of 68 degrees, as depicted in Fig. 2.4. The incident angle can be slightly adjusted from 63 to 70 degrees. As can be seen from the experimental scheme, the non-polarized incident light from a halogen lamp (Avantes, AvaLight-HPLED) passes a linear polarizer if scattering in a specific polarization is interesting. Then the light is focused on the specific nanoparticle by a Mitutoyo Apochromatic Objective with 10 times of magnification (MY10X-803, NA = 0.28). The scattered light is collected by another Mitutoyo Apochromatic Objective with 50 times (MY50X-805, NA=0.55) of magnification. Due to the large oblique incident angle, the reflection angle is higher than the collecting angle of the top objective and the reflection signal is excluded. The collected scattering signal passes the pinhole and directs to the spectrometer (HORIBA LabRam HR). The spectrometer is equipped with one Si detector for visible range and one InGaAs detector for NIR measurements. A cooled charge-coupled device (CCD) Camera (Andor DU 420A-OE 325) and 150 g/mm diffraction grating are also included.

It should be noted that by adding a quarter-wave plate after the linear polarizer, it is possible to achieve circular polarized excitation light. Since circular scattering is not in this dissertation work, it is not demonstrated in the experimental scheme. Moreover, the optical fiber connected to the halogen

lamp and optical setup should be chosen according to the interesting range. If the resonance of the nanoparticle is close to the UV range, it is highly recommended to use UV-VIS fiber. If the resonance of the nanoparticle is adjacent to the near-infrared (NIR) range, VIS-NIR fiber is much more proper. In any case, a bigger diameter of fiber can focus much more light on the nanoparticle and thus receive the higher intensity of the scattering light. This interesting range principle also operates in the selection of collected objectives. DF measurements of nanostructures are the fundamental understanding of the resonant properties of the hybrid nanostructures. The results are included in chapter 3 and chapter 4.

2.2.2 Raman scattering

Above section demonstrates Mie scattering, one type of elastic scattering, which gives information about how the nanoparticles scatter light in free space and without modifying the light frequency. In a contrast, inelastic scattering changes the frequency of incident photon and interact inside the medium. To achieve inelastic scattering in the crystal, additional collective excitations should be involved, for instance, phonons, magnons, plasmons, and so on. Phonons are the common quasiparticles involved in inelastic scattering. According to the different types of phonons involved, one can distinguish Raman scattering (with optical phonons) and Brillouin scattering (with acoustic phonons). These two processes are physically identical but should be analysed by different experimental setups, here only Raman scattering technique is introduced due to the relation to this dissertation work.

As inelastic scattering changes the light frequency, the difference of the photon energy is related with the phonons (or other types of excitations) absorption or emission in the medium, i.e., the medium absorbs phonons if the scattering photon energy increases (Anti-Stokes scattering) and the medium emits phonons if the scattering photon energy decreases (Stokes scattering). That is to say, Stokes scattering is able to occur at any temperature while Anti-Stokes scattering can only be realized with existing phonons before light excitation, resulting in a low possibility at cryogenic temperature. Besides,

inelastic light scattering is generally weak processes and the detection requires a powerful source (for example, laser) and very sensitive detectors.

Glass

Figure 2.5 — Schematic view of confocal Raman spectroscopy

Raman scattering denotes to inelastic scattering interact with optical phonons, which is essentially dispersionless adjacent to the area wave vector equals to zero. This technique is mainly utilized to determine the longitudinal optic lattice vibrations (LO) and transverse optical lattice vibration (TO) modes near the Brillouin zone center in crystals. According to the rule of mutual exclusion, for centrosymmetric structure materials, LO mode could be Raman active and then TO mode can only be IR active, or vice versa. However, TO/LO mode can be both Raman active in non-centrosymmetric structures. That explains why centrosymmetric material (Si, Ge) has only one peak while non-centrosymmetric material (GaAs, InP, GaP, AlSb) have two TO/LO peaks (LO mode locates at a higher frequency) in Raman spectra (depicted against the wavenumber shift: 1 cm-1 equals to an energy shift of

0.124 meV).

In this dissertation work, a confocal Raman microscopy is used to detect the Raman scattering, as plotted in Fig. 2.5. For the excitation source, a HeNe laser (Thorlabs) beam centered at 632.8 nm is applied to the system. As the laser intensity can be as high as 10 mW, a couple of neutral density (ND) filters are applied to flexibly tune the laser intensity for alignment or Raman measurements at different intensities. Subsequently, the laser beam passes a bandpass filter (Thorlabs, FL632.8-10) and be focused by a Mitutoyo Apochromatic Objective with 100 magnification times (MY100X NA = 0.9) placed on the top of the sample. The laser spot diameter can be estimated by Diaser = 1.22 x X/NA. The Raman scattering and reflected excitation beam are collected by the same objective and propagate through a notch filter (Thorlabs, NF633-25) to eliminate the excitation light. Finally, the Raman signal passes the pinhole and directs to the same LabRam spectrometer as DF measurements. Detailed experimental analysis is described in chapter 3.

2.2.3 Scanning near-field optical microscopy

In order to study the near-field properties of the hybrid sample, aperture-type scanning near-field optical microscopy (SNOM) is applied. SNOM is fundamentally different from traditional optical microscopy due to its possibility to map the near-field distribution with subwavelength resolution beyond the diffraction limit, which is 1 to 2 orders higher than that achieved from a typical optical microscope [56]. Therefore, during the last several decades, SNOM has gradually turned into one important and efficient tool for demonstrating the electromagnetic field adjacent to the nanostructures.

In this dissertation work, aperture-type SNOM (AIST-NT) in transmission configuration with a near-field probe is applied. The experimental scheme is presented in Fig. 2.6. As can be seen from the figure, Supercontin-uum laser source Fianinum WhiteLase SC400-6 is utilized as the excitation source, equipped with a tunable Fianium SuperChrome filter. Then a tunable central wavelength (400-850 nm) laser beam with a spectral width of 10 nm and moderate laser power 2-4 mW in the visible range is applied to the

Figure 2.6 — Schematic view of aperture type scanning near-field optical microscopy in

transmission geometry

system. Such low excitation power can not induce shape reconfiguration of the hybrid nanostructures and thus the near-field mapping process doesn't affect the fs-laser modification result. The laser beam passes the linear polarizer and the polarization is along the Y-axis (three oligomers center line, see inset). Then a Mitutoyo objective with 20 magnification is used to focus the excitation beam. To characterize the near-field of the complex oligomers, mapping near-field is conducted with a piezo stage controller utilizing polymer aperture-type probe scanning in constant-height mode. Control system has two functionalities: first, precisely control the probe scanning over the sample surface two-dimensionally; second, once the probe encounters fluctuations of the sample surface during scanning, then the control system will dominate the probe to move up or down according to the changing signal to keep the distance between the probe and the sample surface constant. In this case, SNOM is able to obtain the near-field information of the nanostructures with a stable distance (120 nm) above the sample. The resolution of the measured optical signal is determined by the probe diameter (aperture size <100 nm), probe scanning displacement accuracy and the control accuracy of the distance between the probe and sample surface instead of the excitation wavelength.

So far, the resolution of the structure can reach nearly 10 nm, i.e., nanoscale detection, and it dramatically exceeds the resolution limit of traditional optical microscopy (^200 nm). After the probe scans the near-field of the hybrid oligomers, the signal is recorded by a photoelectron multiplier (PMT) through the aperture. It should be noted that in order to improve the signal-to-noise ratio, an optical chopper with a modulation frequency up to 5 kHz is placed in the excitation channel after the linear polarizer. In this case, the optical chopper demodulates the collected signal at the reference frequency by a synchronous amplifier. Here the inset shows one of the near-field maps of the hybrid oligomers obtained by this SNOM setup. In Chapter 5, the detailed discussions of experimental near-field reconfigurations of the hybrid oligomers are presented.

2.2.4 Second harmonic generation

Above light-matter interactions in section 2.2.1-2.2.3 are mainly excited with low-intensity light. If the material is excited by very high intensity of the laser beam, the nonlinear process is able to achieve when the light propagates through the material. Second-harmonic generation (SHG) is the most typical nonlinear behaviour studied in nanophotonics and it's not difficult to detect the fundamental wavelength and the generated harmonic signal by one spectrometer. SHG usually can not be obtained from materials with inversion symmetry, such as silicon. Besides, a fs-laser is sufficiently induced to modify the optical properties of the materials, demonstrating an efficient light generated at the double frequency of the applied laser light. However, the interface between crystalline silicon and its oxide layer could act as a sensitive probe for SHG and charge transfer [57,58], which is widely observed with multi-materials [59,60]. Despite the intrinsic high-efficiency of SHG benefited by the low-loss and resonant dielectric materials, electric field induced second harmonic generation(EFISH) has attracted intense attention ascribed to the modulation of second-order susceptibility x(2) by introducing the interaction component of third-order susceptibility with an external bias [28,61,62]. However, nanoscale dynamic control requires high voltages which is unsustainable

at the high operating frequency. Therefore an exciting alternative to utilize EFISH in an optical way consists of hot carrier generation by plasmonic nanostructures and transferring electrons to dielectric systems to formulate the static electric field [63].

Figure 2.7 — Schematic of dark-field scattering and SHG

In this dissertation work, a confocal transmission scheme of experimental SHG configuration is applied, as shown in Fig. 2.7. To study the photon up-conversion of the hybrid nanoparticle, a commercial Yb-doped solid-state ultra-fast laser centered at 1050 nm is applied as the excitation source. The laser pulse duration is 150 fs with 80 MHz repetition rate. The intensity of the laser beam is controlled by an attenuator comprised of two Glan prisms (Glan1, Glan2) and a super achromatic half-wave plate (Thorlabs, 600-2700 nm). The principle of the attenuator is ascribed to the fact that the polarization of the laser beam is selected by the first Glan prism (Glan1), then after passing the half-wave plate, the polarization of the laser beam rotates for a certain angle, and thus it can not fully transit the second linear polarizer (Glan2), resulting in the attenuation of the beam intensity. Then a power meter is utilized to measure the laser intensity excited to the system. To perform a pure excitation at 1050 nm and filter the green light, the laser beam

passes through a long-pass filter (FEL1000, Thorlabs) and then is focused on the hybrid nanoparticle from the substrate side by a Mitutoyo Apochromatic NIR Objective with 10 times magnification (MY10X-823, Thorlabs). The generated SHG signal is collected by a Mitutoyo Apochromatic NIR Objective with 50 times magnification (MY50X-825, Thorlabs) and filter the excitation beam by a short-pass filter (FES1000). It should be noted that a longer cut-off wavelength of the short pass filter is more preferable because silicon is able to generate white light in the visible range in addition to generate second-harmonic waves.

Different materials have various crystalline phase and lead to colorful SHG polarization pattern. To characterize this phenomenon, a third glan prism is inserted at the detection channel to map the polarization pattern by rotating the transmission line of the glan prism. At last, the SHG signal passes through the pinhole and directs to the spectrometer (HORIBA LabRam HR). It should be noted that SHG and DF scattering can be measured simultaneously with our confocal setup. By checking the DF spectra of the hybrid NP each time after SHG measurement, information about the morphology and material properties changing can be presented. Thus it is significant to measure SHG behavior accompanied by DF scattering. Detailed nonlinear results are summarized in chapter 3 of this thesis.

2.3 Numerical simulations

The optical properties (S-parameters, far-field, near-field, etc.) of hybrid nanostructures are numerically simulated by means of CST Microwave Studio for high-frequency simulation. CST Microwave Studio is an effective way to explore the electromagnetic properties by solving Maxwell's equation for arbitrary morphology with proper boundary condition. Three typical solvers including time-domain solver, frequency-domain solver and integral equation solver are applied to consider different geometries, resonant properties, substrates and so on. Time-domain solver includes transient solver based on Finite Integration Technique (FIT) and TLM solver utilizing Transmission line Matrix (TLM) and time-domain solver works only with hexahedral mesh

in combination with Perfect Boundary Approximation (PBA) technique. The transient solver is suited for typical nanoantennas calculation and is able to obtain broadband far-field and S-parameters for one calculation run, thus the time-domain solver is characterized as a remarkably efficient solver.

As for the frequency-domain solver, Maxwell's equations are transformed into the frequency domain if the excitation is time-harmonic dependent on the fields. Maxwell equations are solved based on the finite element technique in frequency domain solver. Unlike time domain solver, the frequency-domain solver can only perform the calculations frequency by frequency, each frequency simulation requires solving an equation system. Thus frequency-domain solver mostly suits for small amounts of frequency points. On the other hand, the frequency-domain solver supports both tetrahedral and tetrahedral mesh contributing to solving multi-problems. Besides, powerful post-processing with frequency-domain solver provides solutions to calculate scattering on different substrates with different collecting angles and circularly polarized light excitation which are unavailable for the time-domain solver. Thus frequency-domain solver is my favourite solver even sometimes it's time-consuming.

Integral equation solver is related to surface mesh and also can only be simulated frequency by frequency. An integral equation solver can calculate the total scattering of nanoantennas on different substrates directly without post-processing. Specified angle collection can be realized with postprocessing. However, due to the linear disperse calculation area on the object boundaries instead of volume division, the integral equation solver is not able to calculate the fields inside the nanosystems and exclusively characterize the fields near the objects.

In summary, all of the above three solvers are very competitive and flexible to switch on demand. As for this dissertation research, I usually start with the time-domain server to fast calculate and select out the approximate parameters for nanoantennas in free space. For more specific and accurate fabrication, I transform to integral equation solver and frequency domain solver according to the experimental setups and substrates.

2.4 Chapter conclusions

This chapter gives a basic understanding of different fabrication methods, experimental setups for linear and nonlinear properties exploring and numerical method for simulating the optical properties of nanostructures. The position precision, repeatability, simplicity features are compared for lithography and femtosecond laser treatment fabrication methods. Besides, the excitation sources, objectives, optical elements such as filters, mirrors, fibers, spectrometers of the experimental scheme are comprehensively introduced. Finally, the advantages and limitations in different solvers of CST Microwave Studio are compared for numerical simulations.

Nanophotonics attracts researchers from different fields due to the strong interactions between light and matter in nanoscale. As is well known, significant and efficient confinement of the electromagnetic field contributes to field intensity enhancement, i.e., light-matter interaction strengthening. Typically two ways are proposed to achieve the field confinement: Mie-resonances enhancement in dielectric nanostructures and plasmonic confinement in the metal nanostructure. In a hybrid metal/dielectric nanoantenna, it is possible to combine these two features and control near-field after lithography fabrication via fs-laser irradiation.

By applying different laser intensity and repetition rate, different shapes of gold components can be observed with the gold-silicon nanoantenna and thus provide tuning of near-field distribution. Chapter 3 studies the reshaping that occurred at pulsed fs-laser irradiation with 50 Hz repetition rate. A sphere-cone gold nanoantenna with a certain gap is realized. Thus the role of the distance between two components in near-field control is investigate and applied to study single-photon emission and nonlinear enhancement. Chapter 4 and Chapter 5 is related to fs-laser reshaping at a high repetition rate (80 MHz). Different shapes (disk, cup and sphere) of the gold component are precisely achieved at different laser intensities. In chapter 4 numerical simulations are performed to study the shape modification induced LSPR shift and near-field reconfiguration in a single and pairs of gold-silicon nanoantenna as well as the distance influence between the pairs of hybrid nanoantenna. Chapter 5

demonstrates the near-field reconfiguration by SNOM in hybrid heptamer as the shape of gold component changes by fs-laser irradiation.

CHAPTER 3. FIELD CONFINEMENT IN GOLD-SILICON NANOANTENNAS

This chapter studies the distance dependence of field confinement in a hybrid gold-silicon nanoantenna within a certain gap induced by fs-laser irradiation at a low repetition rate. Section 3.1 demonstrates the electric field enhancement in the gold-silicon nanoantenna as a function of the gap and provides a high potential for strengthening normalized local density of states (Purcell factor). Section. 3.2 shows the giant Purcell factor in the resonant hybrid nanoantenna as expected and summarizes the solid-state single-photon emitters for integration. This type of sphere-gap-cone nanoantenna is fabricated by fs-laser reshaping with a combination of the lithography process. Section. 3.3 presents the electric field enhanced second harmonic generation from the sphere-gap-cone nanoantenna.

3.1 Strong field localization in gold-silicon nanoantenna

The configuration of the hybrid gold-silicon nanoantenna is schematically presented in the inset of Fig. 3.1. The truncated silicon nanocone has a bottom diameter and height of 190 nm. The upper diameter of the nanocone is chosen as half of the bottom diameter due to the easy fabrication process by chemical etching. It exhibits the magnetic dipole resonance and electric dipole resonance in the visible range. Thus Mie-resonances induced electric field enhancement can be realized in the dielectric part. The gold nanosphere has a rare contribution of resonance in the long-wavelength range, so the diameter of the nanosphere is defined as the upper diameter of the nanocone (i.e., 95 nm) for a stable system. The gap between the silicon nanocone and gold nanosphere varies at the range of 10-45 nm, corresponding to the size of the studied quantum emitters in the next section.

To study the electric near-field enhancement (^enhancement) in the center of the gold-silicon nanoantenna, numerical simulations are performed to calculate the ^enhancement as a function of the gap and operating wavelength in CST Microwave Studio. Here an E-probe is inserted for quantita-

10 15 20 25 30 35 40 45 550 600 650 700

gap[nm] wavelength[nm]

Figure 3.1 — Electric near-field enhancement of the hybrid metal-dielectric nanoantenna:

(a) A = 637 nm, ^enhancement dependence on the gap between gold and silicon components under TE/TM polarization; (b) For TM-polarized excitation, ^enhancement dependence on the gap and operating wavelength in the long-range

tively identifying the E .enhancement. The electromagnetic wave is oblique incident upon the gold-silicon nanoantenna at 68 degrees along its symmetry axis, as indicated in Fig. 3.1a. Therefore the distance dependence of the electric field enhancement at the resonance is demonstrated. As the strong electric field confinement is promising for nanoscale sources for PL enhancement, among them nanodiamond is quite interesting and commercially available. Thus preliminary calculation are performed to obtain the resonances (518 nm and 637 nm) by putting a negatively charged nitrogen-vacancy (NV-) nanodiamond in the gap of gold-silicon nanoantenna. It should be noted that Fig. 3.1 only shows the calculated results at 637 nm due to the spectrum at 518 nm is almost overlapped.

As can be seen from the figure, since the electric vector oscillates in the incident plane exciting the electric resonances for TM-polarization efficiently and thus strongly localizes the electric energy in the gap, the electric field enhancement is dramatically higher than that under TE-polarized excitation. Moreover, as the gap increases, the field confinement gradually decreases for TM-polarization and keeps depressed for TE-polarization. Besides, Fig. 3.1b plots the operating wavelength and gap dependence for TM-polarized excita-

H

H

#0

Figure 3.2 — E-field and H-field distribution for the hybrid gold-silicon nanoantenna under TE- and TM-polarization excitation: (a) A = 637 nm; (b) A = 518 nm

tion. It clearly shows the optimal strategies for near-field control: (i) electric field localization by a plasmonic nanosphere; (2) electric field confinement at the resonance raised from the dielectric nanocone coupled with a nanodia-mond.

To qualitatively describe the electric field enhancement, electric field (E-field) and magnetic field (H-field) distributions at 637 nm and 518 nm under TE/TM polarization with 10 nm gap are plotted in Fig. 3.2. For TM-polarized excitation, incident light is strongly localized in the gap for both wavelengths. The slight difference is the enhancement at 637 nm originates from magnetic resonance while the electric quadruple resonance leads to the enhancement at 518 nm. Two folds enhanced field confinement in the hybrid gold-silicon nanoantenna provides high potential in effective pumping single-photon emitters such as nanodiamonds and quantum dots which is investigated in the next section.

3.2 Field enhanced Purcell factor

As is well-known, a two-level system should decay spontaneously during the interaction with a vacuum media at a certain rate proportional to the spectral density of modes per volume which is evaluated at the transition frequency. Thus the density of modes in the cavity will change, and the

amplitude will fluctuate greatly. From the point of view of cavity mode (it must be regarded as a quasi-mode in the presence of dissipation), the maximum density of the mode appears at the quasi-mode resonance frequency and may strongly exceed the corresponding density in free space [64]. In 1940s, Purcell came to this conclusion by pointing out that a single (quasi) mode occupies a spectral bandwidth n/Q in the cavity of volume V. By normalizing the obtained cavity enhancement mode density per unit volume to the free space mode density, it is defined that the modified spontaneous emission rate induced by the cavity as Purcell factor, which can be expressed as [65]:

F =

A ( * Q

4^2 ( n) V

(3.1)

where A is the wavelength in free space and n is the refractive index of cavity material. Q is the quality factor within the cavity and V is the mode volume of the cavity. So when the atom transition locates within the mode linewidth, the Purcell factor actively increases the spontaneous decay rate.

Figure 3.3 — Illustration of Purcell effect: (a) spontaneous emission driven by local density of modes in free space; (b) coupling to a cavity and Purcell enhanced spontaneous rate of emission by increasing the density of modes. Adopted from [64]

The illustration of the Purcell effect is plotted in Fig. 3.3. For achieving a high Purcell factor in micro-/nanocavity (closed or open cavity), the cavity design should consider the corresponding atomic transition characteristics. It

means the local density modes modification is more significant if the cavity is a resonator and its resonance is tuned to the emission frequency of the source. Even though the Purcell effect was first demonstrated in circuits with nuclear magnetic resonance [66], During the last several decades, researchers have applied this concept to many other applications, such as enhanced light-emitting devices in microcavity [67], plasmonic nanoantenna enhanced single-molecule emission [68], as well as tailoring optical nonlinearities via Purcell factor [69].

According to Eq. 3.1, high quality factor resonance within a small volume cavity contributes to high value of Purcell factor. In fact, manipulating the Q factor alone is restricted by the transition spectrum width. However, if all the other parameters are set, the Purcell transition can be strengthened more narrowly as Q factor increases relatively higher. So small quantum emitters such as QDs, color center nanodiamond and perovskites with narrower transition width possess high potential in photoluminesence applications in nanoscale.

Table 3.1 — Solid State single-photon emitters

Materials Types Band gap, eV ZPL 9(2)(0) T Ref.

hBN insulator — 6 - 560-800 nm 0.077 RT [70,71]

TMDC semi-C - 1.0 - 2.5 - 600-900 nm 0.14 ± 0.04 CT [72,73]

Graphene semimetal zero-gap — 650 nm ¡0.1 RT [74,75]

CNT semi-C 1 eV for dt « 1 nm - 0.85-2 yU,m 0.0028 ± 0.0012 0.01 ± 0.01 CT RT [76, 77]

QD semi-C vary with its size - 0.3-2.0 yU,m As QDs i 0.01 N QDs - 0.3 CT RT [78, 79]

Colour centers in diamond semi-C - 5.5 NV- : 637 nm NV0 : 575 nm SiV- : 738 nm NV: - 0.32 SiV: - 0.16 RT [80,81]

Colour centers SiC: 3-4 - 675-700 nm

in compound semi-C YAG: 6.3-6.5 Pr3+ : 300-450 nm - 0.1 RT [82-85]

semi-C ZnO: 3.4 - 560-720 nm

Perovskite semi-C - 1.5-3 - 400-800 nm CsPbl3 : 0.06 CT & RT [86, 87]

* TMDC: Two-dimensional transition metal dichalcogenide; CT: cryogenic temperatures; semi-C: semiconductor; As QDs: Arsenide QDs; N QDs: Nitride QDs

There are plenty of candidates for solid-state single-photon emitters

and the size alters from several atom layers to tens of nanometers, suiting for the distance between the gold sphere and silicon cone studied in the last section. By changing the quantum sources, the emission wavelength can cover a wide range from ultraviolet to infrared pumped by a green laser centered at 532 nm. To lay the basis for future research with different solid-state singlephoton emitters, the optical properties for the promising quantum sources are summarized in Table 3.1.

Materials cover various 2D materials (hexagonal Boron Nitride (hBN), two-dimensional transition metal dichalcogenide; (TMDC), graphene, carbon nanotube (CNT)), quantum dots (QDs), color centers in diamonds and compound semiconductor (semi-C) and perovskites. Optical properties include conductor types, band gaps, zero-phonon-line (ZPL) wavelength, second-order correlation function (g(2)(0)) and temperature conditions. Here ZPL wavelength indicates the energy difference between the excited state and ground state since the excitation/relaxation is not phonon-assisted, which is very important to study the photoluminesence of the quantum sources. g(2)(0) is the reference parameter to indicate the single-photon purity, i.e., g(2)(0) < 0.5 corresponds to a single photon source and g(2)(0) > 0.5 identifies a multiple photon emission.

The proposed gold-silicon nanoantenna exhibits strong electric field confinement in the gap with distance dependence, which can be applied to couple with different quantum emitters for enhanced emission, as schematically shown in Fig. 3.4 taking a nanodiamond as an example.

Since the single-photon emitters are coupled with the proposed goldsilicon nanoantenna, the spontaneous emission rate characterized as Purcell factor is supposed to be strongly modified. Since the nanodiamonds with color centers have a wide range of dimensions and are already commercially avail-able,numerical simulations are performed to calculate the Purcell factor with a nanodiamond occupied the gap of gold-silicon nanoantenna on a glass substrate, as shown in Fig. 3.4. The refractive index n of the nanodiamond is 2.4 without losses. The quantum source is simplified as a discrete electric dipole (length = 5 nm). Since the emitter orientation in the nanodiamond is ran-

o

Figure 3.4 — Schematic show of the gold-silicon nanoantenna integrated with quantum emitters (for instance, a nanodiamond). The system with different quantum sources is excited by a green laser centered at 532 nm and emit photoluminescence signal from

ultraviolet to infrared wavelength range

dom, the Purcell factor is calculated with a horizontal and vertical orientated dipole, respectively.

As mentioned before, the Purcell factor is the normalized local density of states (LDOS) and it can be evaluated by considering the imaginary part of the Green's function of a point dipole source which is located at the point ro on itself [88], which identifies as Im[Gzz(ro,ro,u)]. Here u is the radiation frequency of the dipole source. Thus the Purcell factor can be calculated in the stationary regime by considering the input impedance of an electric dipole source and its radiated power, which can be expressed by Green's function as [88,89]:

= Im[Gzz(ro,ro, u)] = Re[Zm] (3 2)

= Im[G°zz(7-0,7-0, u)] = Re[Zo,m] ( . )

where Re[Z^n] and Re[Zo,in] indicate the real part of the input impedance of the electric dipole coupled to the gold-silicon nanoantenna and in free space, respectively. These two values can be numerically calculated in CST Microwave Studio. For precise calculation, the same mesh settings are obligatory to keep the same for the dipole source with and without the gold-silicon nanoantenna.

Similar to Gzz, G°zz indicates the Green's function in free space corresponding to the same point. According to Eq. 3.2, as Re[Zz,in] is a constant value for a specific emission source at a certain frequency, the value of Re[Z^n] takes the majority role in the Purcell factor values.

400 500 600 400 500 600

wavelength[nm] wavelength[nm]

Figure 3.5 — Purcell factor in log-scale achieved in the gold-silicon nanoantenna coupled

with a nanodiamond. (a) horizontal dipole orientation; (b) vertical dipole orientation

Fig. 3.5 depicts the total Purcell factor in log-scale achieved in the hybrid nanoantenna coupled with nanodiamonds with variable size (10-45 nm) in the visible range for two dipole orientations. As can be seen from the figure, no matter the dipole orientates along horizontally or vertically, the resonances in the hybrid system are both centered at 518 nm and 637 nm, contributing to as high as 103 and 103 8 for horizontal and vertical dipole orientation, respectively. This again proves the concept that a combination of resonant dielectrics and plasmonic nanoparticles results in strong electric field enhancement and thus strengthening the Purcell factor. In fact, this size of hybrid nanoantenna is designed for integrating nanodiamonds with negatively charged nitrogen-vacancy (NV-) color center, the high Q resonance (A = 637 nm) is exactly the ZPL wavelength of this type of single-photon emitter (see Table 3.1), which is supposed to strongly enhance the singlephoton emission of the quantum emitter. However, this hybrid nanoantenna is not limited to applications with nanodiamonds. By carefully choosing the geometry for the silicon nanocone and gold nanosphere, the resonance can cover the whole wavelength range from ultraviolet to infrared and suitable for

coupling with other single-photon sources in Table 3.1. For instance, even with the current geometry, another broadband resonance centered at 518 nm has a high potential for enhancing the emission from hBN and QDs.

As is well know, only the radiative Purcell factor contributes to efficiently strengthening photoluminesence (PL) while the non-radiative part indicates the losses. In this case, further calculations are conducted to study the radiative power and radiative Purcell factor for horizontal and vertical dipole orientation. On the other side, since the dipole orientation in the nan-odiamond is randomly distributed, further the average radiative Purcell factor is calculated by:

Faverage 2/3^kor + 1 /'3Fver (3.3)

where F^or and Fver represent the Purcell factor with horizontal and vertical dipole orientation, respectively. Proportion is ascribed to the degrees of freedom in three-dimensional space.

25 30 gap[nm]

Figure 3.6 — Average radiative Purcell factor as a function of the gap (nanodiamond size)

at 518 nm and 637 nm

Fig. 3.6 depicts the average radiative Purcell factor for the hybrid goldsilicon nanoantenna coupling with a NV- nanodiamond at 518 nm and 637

nm. Both of them gradually decrease with the gap increases, corresponding to the electric-field enhancement decreasing in Fig. 3.1. Even it is one order lower than the total Purcell factor, but it's still high and quite competitive for dielectric or plasmonic systems [18,23-25].

180 ° 180 °

A, = 637 nm X = 637 nm

Figure 3.7 — Radiation pattern for (a) horizontal dipole orientation at 518 nm; (b) vertical

dipole orientation at 518 nm; (c) horizontal dipole orientation at 637 nm; (d) vertical dipole

orientation at 637 nm

A hybrid resonant metal-dielectric cavity not only plays an important role in strong field confinement and enhanced spontaneous emission but also exhibits the ability to tailor the radiation pattern, which is quite important for single-photon emission. Fig. 3.7 shows the radiation pattern of the coupled hybrid system at 518 nm and 637 nm for both dipole orientations. For vertical orientation, it oscillates along the symmetry axis of the hybrid nanoantenna and represents relatively similar directivity, i.e., two lobes perpendicular to the symmetry axis. However, for horizontal dipole orientation, two resonances present dramatically different radiation patterns. At 518 nm the nanodiamond

emits unidirectional perpendicular to the emission direction which is promising for collecting all the emission signal from the top. On the other hand, at 637 nm the nanodiamond emits forward and backward identically along the dipole emission direction. This indicates the benefits from hybrid metal-dielectric systems by combing the advantages from plasmonics and dielectrics.

As discussed above, the proposed gold-silicon nanoantenna demonstrates strong electric field enhancement with induced high Purcell factor, radiation directivity control. However, it is necessary to compare the performance with plasmonic and dielectric systems. Plasmonic nanoparticles can strongly localize the electric field within the vicinity of the nanostructures [46]. Nevertheless, it is difficult for most plasmonic nanostructures, for instance, spheres and disks to control the radiation directivity due to the monotonous electric resonances. In contrast to plasmonic nanoparticles, the dielectric ones usually manipulate radiation direction via coupling magnetic and electric resonances [23,24]. Thus hybrid metal-dielectric systems are proposed to sum up the advantages in plasmonic and dielectric materials [18,25]. Therefore the studied optical properties of reported hybrid metal-dielectric, plasmonic and dielectric systems are summarized to make a comparison with our proposed gold-silicon nanoantenna coupled with 10 nm nanodiamond in table 3.2.

Table 3.2 — Comparison of various dielectric and metallic nanosystems

performance

Nanosystems Directivity Purcell factor Field enhancement Ref.

Plasmonic cavity - 3.5 x 106 - [46]

Dielectric Yagi-Uda 12 4 - [23]

Dielectric notch antenna 10 - - [24]

Hybrid Janus dimer - - 120 [18]

Hybrid Yagi-Uda antenna 49 1800 - [25]

* Proposed hybrid nanoantenna 2 103.8 11 -

* Performance of the proposed hybrid nanoantenna calculated at gap = 10 nm

To conclude, strong electric field enhancement at the broadband Purcell factor resonance with high directivity provide the proposed hybrid nanostruc-ture good chances for effective controlling and enhancing the single-photon emission.

3.3 Influence of E-field enhancement on SHG

As discussed in section 3.1, the proposed gold-silicon nanoantenna exhibits strong electric field enhancement in the gap. By integrating the nanoantenna with a quantum source, a high Purcell factor is achieved which is promising for photoluminesence enhancement. In this case, this electric field localization can be applied to enhance the nonlinear behavior of the gold-silicon nanoantenna. Similarly, by placing the dielectric nanowires in the hot spots of plasmonic oligomers, efficient second-harmonic generation is demonstrated [51]. Thus the proposed hybrid gold-silicon nanoantenna with a certain gap is fabricated by pulsed fs-laser reshaping and the electric field enhanced second-harmonic generation behavior is investigated expecting one order enhancement compared to the corresponding silicon nanocone.

Figure 3.8 — SEM images of the sphere-cone-gap nanoantenna reshaping process with laser

power density increasing: (a) original shape, gold nanodisk on silicon nanocone; (b) gold nanosphere on silicon nanocone without gap; (c) desired sphere-gap-cone nanoantenna; (d)

gold nanosphere drops off the silicon nanocone; (e) sphere jumps away, silicon nanocone remains presumably capturing some part of Cr named as R-cone. All scale bars are 200 nm

The fabrication process of the original disk-cone nanoantenna (see Fig. 3.8a) combines e-beam lithography, Au and Cr evaporation, lift-off procedure, gas-phase chemical etching (details see lithography fabrication in section 2.1). The silicon nanocone has the same geometry parameter in the above studies. Here a thin Cr layer (^ 2 nm) is deposited for strong adhesion of silicon and gold materials. Besides, due to its relatively high melting tempera-

ture comparing to gold, when the fs-laser pulse irradiates on the structure, the Cr stretches and covers the gold particle to keep the gold volume unchanged in accordance to the studies for the continuous wave modification process [26]. In fact, this thin Cr is a major role in the gap between the gold nanosphere and silicon nanocone.

A commercial Ti:Sapphire laser centered at 800 nm (Avesta project, Tpuise ~ 100 fs, repetition rate = 80 MHz)) combined with motorized linear translators with air suspension (Aerotech Inc., USA) is utilized to modify the shape of the gold component from nanodisk to nanospheres. The fs-laser is focused by an Olympus Objective with 40 x magnification (numerical aperture (NA) = 0.75) and irradiated on the nanostructure from the top. Thus the diameter of the laser spot can be calculated by:

1.22 x A

Ulaser = NA (3.4)

where A is the wavelength of the laser beam and NA is the numerical aperture of the focused objective. As a result, the laser spot is approximately 1.3 ym. Since the distance between each gold-silicon nanoparticle is around 1 ym, so each nanoparticle can be precisely modified by fs-laser irradiation without affecting the neighbouring nanoparticle.

As can be seen from Fig. 3.8, with the laser power density increasing, the gold nanoparticle undergoes dramatic shape transformation. Conversion to sphere-gap-cone nanoantenna (gap nanoantenna) occurs at 845 GW/cm2 with 50 Hz repetition rate. It should be noted that a low repetition rate is obligated. Since the sample is moved at a speed of 10 ym with the help of the piezo stage, and the distance between each gold-silicon nanoparticle is around 1 ym. In this case, a low repetition rate contributes to single pulse irradiation for one hybrid nanoparticle and results in gap nanoantenna configuration (see Fig. 3.8c). This slight detachment of gold nanoparticle from the silicon nanocone can be explained by the transformation of surface energy to kinetic energy and thus lift the center of mass of the gold nanoparticle, which is well-studied in [27] with triangular gold nanoparticles on both graphite and glass substrates. The energy difference can be expressed as:

AE = 7lv((1 - cos0)%r2 + 2nrd - 4<kR2) (3.5)

where 7 iv is the liquid-vapor surface tension of gold (1.15 N/m) [90]; 6 is the contact angle of liquid gold on glass (140 ) [91] ; r is the radius of the gold nanodisk (95 nm) and d is the height of nanodisk (20 nm), R is the radius of the detached gold nanosphere (47.5 nm).

However, as the laser intensity exceeds 845 GW/cm2 to some extent, the gold nanosphere obtains enough energy to jump away from the silicon nanocone and Cr layer resulting in the adjacent nanoparticles (« 950 GW/cm2, see Fig. 3.8d) and silicon nanocone presumably with remaining Cr part, named as R-cone (« 1200 GW/cm2, see Fig. 3.8e). Thus the performance of gap nanoantenna and R-cone can be compared with each other during the nonlinear experiments.

0.9 0.8 -0.7

I 0.6

0.5

VJ

£ 0.4 0.3 0.2 0.1 0

400 450 500 550 600 Raman shift (cm1)

Figure 3.9 — Raman spectra. Blue spectrum labelled as before is measured from the initial gold-disk nanoantenna, red spectrum labelled after is measured from gap nanoantenna and

R-cone after fs-laser reshaping

Apart from the shape modification of the gold component, the silicon nanocone has crystal structure transformation instead of shape reconfiguration

due to the laser-induced heating temperature is higher than the silicon crystallization temperature (650 °C) but lower than silicon melting temperature (1687 °C) [92,93]. This is confirmed by Raman measurements in Fig. 3.9. Before fs-laser reshaping, lithography fabricated silicon nanocone is composed of amorphous silicon (a-Si) and presents a broadband a-Si Raman shift peak centered at 480 cm-1. However, after the fs-laser modification, a polycrystalline state is observed in the gap nanoantenna and R-cone which exhibit a much narrower crystalline peak at 521.5 cm-1. As is well-known, crystalline silicon nanostructures provide local inversion symmetry breaking at the interface and a high possibility of achieving efficient second-harmonic generation [94].

Figure 3.10 — Scattering spectra of gap nanoantenna and R-cone: (a) bottom diameter and height of the cone are both 190 nm; (b) bottom diameter and height of the cone are 270 nm

and 110 nm, respectively. (c-d) SEM images of the gap nanoantenna and R-cone in (a); (e-f) SEM images of the gap nanoantenna and R-cone in (b). Scale bars are 200 nm. (g-h) E-field distribution for gap nanoantenna and R-cone of size in (b)

Since the localized electric field can strongly enhance the nonlinear process, it is necessary to keep in mind the key features to achieve strong field confinement: resonant dielectric nanocones and confinement induced by plas-monic nanospheres. In the absence of nanodiamonds,the resonant properties

of the gap nanoantenna and R-cone are recalculated. The nanoparticles are normally excited by an electromagnetic wave from the substrate side (SHG measurement scheme, see Fig. 2.7) in CST Microwave Studio. The scattering spectra for two sizes are presented in Fig. 3.10a-b. As it can be seen from the figure, the original size (D&oi « 190 nm, hcone « 190 nm) has distinct resonances in the visible range but not at the SHG wavelength (525 nm, the excitation laser for the nonlinear experiment is centered at 1050 nm). However, Fig. 3.10b shows pronounced resonances at 525 nm for gap nanoantenna and R-cone with another geometry of the nanocone (D&oi « 270 nm, hcone « 110 nm). Their corresponding SEM images and E-field distributions are presented in Fig. 3.10e-f and Fig. 3.10g-h, respectively. The recalculated size of the gold-silicon nanoparticle has strong electric field localization in the gap (Cr layer) as the initially studied nanostructure.

Figure 3.11 — (a) The maximum SHG intensity obtained in gold-silicon gao nanoantenna, R-cone and 100 nm thickness of Si film, respectively; (b) scattering spectra of gap nanoantenna measured after each SHG detection by confocal dark-field microscopy

Here two benefits are generated from the new geometry of the hybrid nanoantenna: first, it is proved that resonance at the SHG wavelength can increase the SHG intensity and reduce the time of SHG accumulation in silicon-based nanostructures [94]; second, strong localization of electric field

is observed within the gap nanoantenna. Therefore the nonlinear experiments are conducted with the new geometry as well as the performance of the gap nanoantenna with the R-cone which have comparable far-field properties but dramatically differs in the near-field configuration are compared.

Fig. 3.11a demonstrates the SHG spectra generated from the gap nanoantenna, R-cone and 100 nm thickness of silicon film with maximum intensity. Here the nonlinear experiments are carried out in the identical measurement condition (laser focusing, objective, etc.) under the irradiation of fs-laser at 1050 nm for all the samples in transmission scheme. Previous studies show there is no significant differences in SHG signal from different thickness of Si films (50, 100, 120 nm) [94]. The maximum SHG intensity is obtained near the damage threshold of each case which is derived from the modification of scattering spectra. An example of the scattering spectra measured after each increased excitation power for gap nanoantenna is plotted in Fig. 3.11b. The scattering spectrum is dramatically changed after the gap nanoantenna is damaged. Following the same method, the damage threshold for gap nanoantenna, R-cone and thin Si film is 284 GW/cm2, 79 GW/cm2 and 35 GW/cm2, respectively. Since the damage threshold for the gap nanoantenna is 3.5 times higher than the R-cone and 8 times higher than the silicon film, one can observe the maximum SHG intensity of gap nanoantenna is 2 orders and 3 orders higher than that from the R-cone and silicon film.

To understand the mechanism for damage threshold difference between the gap nanoantenna and the R-cone, numerical simulations are performed to investigate the fs-laser induced heating properties of these two structures under the continuous wave (CW) laser at the same wavelength (1050 nm) and intensity (70 GW/m2). Since silicon has low loss in the near-infrared region and Cr has relatively low thermal conductivity (93.7 W/mK) compared to gold (317 W/mK) and Si (148 W/mK) [92], the thermal calculations take the Cr layer into account. Moreover, according to the formula in [95], the average heating of the nanostructures in pulsed laser (RR = 80 MHz) excitation in consistency with experimental conditions is estimated. The laser-induced heating properties of the gap nanoantenna and the R-cone are summarized in

Table 3.3 — Fs-laser induced temperature variation in the nanostructures at the wavelength of 1050 nm

Nanostructure CW laser Pulsed laser

Gap nanoantenna 159.67K 128.4197K

R-cone 212.48K 204.9977K

* Laser power density = 70 GW/m2

As can be seen from the table, no matter with CW laser or pulsed laser, the R-cone exhibits higher temperature variation than the gap nanoantenna due to the low thermal conductivity and large absorption in Cr layer. However, in the gap nanoantenna, with the gold sphere located on the top of the cone, laser-induced heating dissipates better. Moreover, with the strong localized electric field, the SHG signal of gap nanoantenna at high damage threshold is 2 orders higher than the R-cone.

Figure 3.12 — Dependence of SHG intensity as a function of the laser power density in log-log scale: (a) gap nanoantenna (blue bricks) and R-cone (yellow triangles); (b) laser-ablated Si sphere (red sphere) for comparison. Inset triangles show the approximation

line in the log-log scale

In fact, the studies in section 3.1 only demonstrate one order improvement from electric field enhancement. To verify the experimental studies,

the dependence of SHG intensity with respect to the pumping fs-laser power densities for the gap nanoantenna and the R-cone are plotted in log-log scale in Fig. 3.12a, demonstrating almost a quadratic slope for R-cone and non-quadratic dependence for the gap nanoantenna. The first assumption for the non-quadratic dependence lies in the incorrect experimental scheme or inappropriate operation skills. Thus the same experiment is conducted with a laser-ablated silicon sphere (fabrication details see section 2.1.2) and show the dependence in Fig. 3.12b, presenting an ideal quadratic slope.

Tracking the silicon-based nonlinear research works, it is not difficult to realize not only the optical electric field but also the static electric field play an important role in the nonlinear experiment of the gap nanoantenna. When a static electric field is exposed to the material, the second-order susceptibility X(2) can be modified by the interactions between the static electric field E^c and the third-order susceptibility x(3), which can be expressed as [28]:

I (2w) = |X(2) + X(3)^C|2/2M (3.6)

Therefore the SHG intensity dependence of the laser intensity is no longer quadratic and this is called electric field enhanced second harmonic generation (EFISH). This effect is widely observed via an external biased voltage with silicon-based systems [28,61], 2D materials [96], polymer [62], ZnO [97]. However, an alternative way to achieve a static electric field lies in multi-photon absorption in the materials and generating electrons, which is also proved in silicon-based systems [57,58,61] and other centrosymmetric materials [60,98].

For pure silicon systems, the potential barrier for Si/SiO2 interface to form the static electric field is 1.55 eV. The previous works use 800 nm fs-laser to pump the silicon-based system to achieve the EFISH effect and modulates the second-order susceptibility. This is also the reason why a purely quadratic dependence is achieved for a silicon nanosphere with the fs-laser centered at 1050 nm as demonstrated in Fig. 3.12b. This is due to the applied fs-laser does not provide enough energy (1.18 eV < 1.55 eV) to form the static electric field. However, for gap nanoantenna, the hot electrons are generated by the gold

nanosphere via multi-photon absorption at the pumping wavelength of 1050 nm and transferred to the silicon nanocone, developing a charge separation at the Si/SiO2 interface. The formed static electric field excites the EFISH effect in the gap nanoantenna with a non-quadratic dependence (~4), as shown in Fig. 3.12a. Moreover, a small amount of Cr layer or Cr/Au mixture left on the silicon nanocone acted as electrons generation source also results in a slight deviation of the slope («2.32). Recently hot carriers generate from plasmonic nanostructures and transfer to dielectric materials are studied by different groups for different purposes, such as nonlinear enhancement [63], photocatalysis [99], and so on.

Figure 3.13 — Polarization pattern for (a) excitation laser at 1050 nm; (b) gap nanoantenna SHG signal at 525 nm; (c) R-cone SHG signal at 525 nm. Solid lines fit from SHG(d) = A • sin2 (9 + 0) and 0 is shown in each figure on the top

At last, Fig. 3.13 shows the polarization pattern for the excitation laser, the gap nanoantenna and the R-cone to demonstrate the exact generation of the second harmonic. This is obtained by inserting a glan prism in the detection channel of SHG experimental scheme (details see section 2.2.4). Since silicon is centrosymmetric material, the SHG polarization patterns have the same shape as the laser radiation. However, since the gap nanoantenna and R-cone have the multigrain structure of silicon component, the polarization pattern is slightly rotated with an angle of 12.7 ° for the gap nanoantenna and 9.1 ° for the R-cone. Comprehensive studies are performed to investigate the polarization-dependent SHG response patterns for all the angle orientation

relative to crystal-axis [100].

3.4 Chapter conclusions

In this chapter, a type of hybrid sphere-cone gold-silicon nanoantenna with a certain gap is proposed to study the near-field control and electric field enhanced applications. Strong electric field localization is observed within the gap of this type of gold-silicon nanoantenna. It can be utilized to integrate with single-photon sources to dramatically enhance the single-photon emission with directivity control. This is verified by Broadband high Purcell factor simulations in section 3.2. When the gold-silicon nanoantenna is coupled with 10 nm nanodiamond, Purcell factor as high as 103 and 1038 for horizontal and vertical dipole orientation is achieved. Even the average radiative Purcell factor can reach 300. Moreover, this type of gold-silicon nanoantenna is fabricated by pulsed fs-laser irradiation and the corresponding nonlinear properties are investigated. It demonstrates not only the optical electric field but also the static electric field contribute to the enhancement of the SHG response. As a result, the gold-silicon gap nanoantenna exhibits 2 and 3 orders higher SHG signal compared to R-cone and thin silicon film. This type of strong field localized gold-silicon nanoantenna is promising for quantum optical chips, field-enhanced spectroscopy, nonlinear sensing and many other applications.

CHAPTER 4. CONTROL OF OPTICAL

FIELD DISTRIBUTION IN THE NEAR-FIELD ZONE THROUGH SHAPE CONFIGURATION OF GOLD-SILICON

NANOANTENNA

Chapter 3 shows the distance between components of gold-silicon nanoantenna that can be used to tune optical field distribution in the near field zone. It is suitable for quantum emitter integration and nonlinear enhancement. Another feature is to reconfigure the shape of the plasmonic component and sequentially tune the near-field confinement. Therefore fs-laser reshaping is proposed to change the shape of the gold component in gold-silicon nanoan-tennas. Such shape modification can be made through irradiation by fs-laser step by step and achieve different contours of gold components (nanodisk, nanocup and nanosphere) [26]. In this chapter, the effect of gold component shape variation in a single gold-silicon nanoantenna is numerically studied and further the coupling effects arising from the interactions of such gold-silicon nanoantennas with reconfigurable gold components are numerically investigated. Section 4.1 demonstrates the modified resonant properties induced scattering, near-field distribution and Purcell factor reconfiguration for a single gold-silicon nanoantenna of three stages. Section 4.2 shows the tuning effect of near-field and far-field properties as well as the Purcell factor for pairs of the gold-silicon nanoparticles.

4.1 Shape induced control of near fields in a single nanoantenna

Inspired by the previous work [26], fs-laser irradiation can be applied to change the shape of gold components in gold-silicon nanoantennas depending on different laser intensities (nanodisk, nanocup and nanosphere). The work starts from exploring the tuning effect of resonant properties by calculating scattering spectra and E-field distributions for a single gold-silicon nanoantenna at different modification stages utilizing CST microwave Studio, as depicted in Fig. 4.1. Since the interesting wavelength lies in the visible

range, the same geometry and materials are utilized as the previous chapter.

Initial hybrid gold-silicon nanoantenna consists of a gold nanodisk located on the top of a truncated silicon nanocone. The diameter of the nanodisk (Ddisk = 190 nm) equals to the bottom diameter of the nanocone (DCone-bottom = 190 nm) due to the fabrication process. The top diameter of the nanocone is half of the bottom diameter, i.e., Dcone-top = 95 nm. The height of the nanocone is = 190 nm. As demonstrated in section 2.1.2, a thin Cr film (« 2 nm) is deposited for adhesion of gold disk and silicon cone, when the gold component changes the shape from nanodisk to nanocup to nanosphere upon fs-laser reshaping, the Cr layer stretches to cover the gold component and keep the volume of the gold part remain the same. On the other hand, since the Cr layer is quite thin, it avoids the influence on the electromagnetic properties of the gold-silicon nanoantenna. Besides, the presence of the Cr layer avoids the forming of Au-Si alloy. Thus the geometry of the gold nanocup and nanosphere can be easily calculated. The nanocup has an outer radius of 80 nm and an inner radius of 62 nm. The radius of the melted nanosphere is 51 nm.

Thus when the electromagnetic light is incident from the top of the hybrid nanostructure, three reshaping stages of the hybrid gold-silicon nanoantenna exhibit dramatically different resonant properties in the visible range. Here only disk-cone, cup-cone and sphere-cone configurations are chosen for studies since the near-field enhancement in further reshaping geometries (sphere-gap-cone nanoantenna and R-cone) are already discussed in chapter 3. The material properties for simulations are adopted from references [101,102]. To achieve a more precise configuration of the near-field at the interface of gold-silicon nanoantenna, a local mesh setting is introduced to improve the calculation accuracy.

Fig. 4.1a depicts the scattering spectra for three stages of the hybrid nanoantenna. The interesting resonant wavelengths of the disk-cone, cup-cone and sphere-cone nanostructures of the scattering spectra (red solid line) are sequentially numbered. Fig. 4.1b depicts the E-field at each corresponding resonance. Moreover, the scattering spectra of the corresponding plasmonic

component in free space is calculated separately and presented in a blue dashed line.

Wavelength i

Figure 4.1 — Investigation of the near-field tuning through the gold component shape of the

hybrid nanoantenna. (a) Scattering spectra for three modification stages depicted in red solid lines, blue lines indicate the scattering spectra from corresponding plasmonic particles in free space; (b) E-field distribution at scattering resonances in (a): resonance 1 at 910 nm, resonance 2 at 680 nm, resonance 3 at 750 nm, resonance 4 at 660 nm, resonance 5 at 680

nm, resonance 6 at 590 nm

For the initial hybrid disk-cone nanoparticle, resonance 1 at 910 nm is the LSPR excited from the gold nanodisk demonstrating strong field intensity enhancement as high as 12. As discussed in section 1.3, the LSPR is quite sensitive to the shape and size of the plasmonic nanostructure. Another two resonances at the wavelengths of 680 nm and 590 nm correspond to the magnetic dipole (MD) and electric dipole (ED) resonance of the silicon nanocone. As is well-known, MD resonance of a dielectric nanosphere is proportional to the diameter and refractive index. When the cone has comparable diameter and height, the MD resonance can be also estimated approximately in the same way. Moreover, around resonance 2, the interaction between the MD resonance from the silicon nanocone and the plasmonic mode from the gold nanodisk appears due to large-area contact interference.

When the gold nanodisk changes to the gold cup, the LSPR of the gold component blue shifts from resonance 1 at 910 nm to resonance 3 at 750

nm. E-field at resonance 3 in Fig. 4.1 proves that and shows a strong electric field enhancement of 9.5. Due to the dramatically decreased contact area, resonance 4 (MD resonance of nanocone) only exhibits the field localization in silicon nanoparticle and the contact edges. By adjusting the geometry size, it is possible to widely tune the hybrid modes in the visible range for specific applications.

With the reshaping process proceeding, the gold nanocup changes to a gold nanosphere. The LSPR of the gold component blue shifts to 590 nm nearly overlapping with the ED resonance of the silicon nanocone and the intensity decreases a lot. E-field distribution at resonance 6 confirms that and shows strong field confinement in the silicon nanocone and adjacent to the gold nanosphere. In fact, the contribution of the nanosphere in the scattering spectrum of gold-silicon nanoantenna can be almost neglected. As a result, the scattering spectra of sphere-cone nanoantenna and nanocone are approximately overlapped. However, hot spots near the gold nanosphere (resonance 6) can be observed, which provides us with a good set of comparisons for nonlinear behavior which is discussed in section 3.3. As a result, the LSPR can be precisely tuned over the visible range by fs-laser reshaping. By carefully design the geometry of plasmonic and dielectric components, the LSPR can overlap with Mie-resonances from dielectrics to achieve broadband enhancement and other effects.

As discussed in chapter 1 and chapter 3, near-field localization strongly affect the normalized local density of states which is characterised by the Purcell factor. Thus the shape variation of the gold-silicon nanoantenna also has an influence on the Purcell factor. Then the Purcell factor for three stages of fs-laser reshaping of the gold-silicon nanoantenna is calculated in CST Microwave Studio.

Fig. 4.2 shows the simulation results of Purcell factor for different shapes of gold components in hybrid nanoantenna from CST Microwave Studio. An electric dipole source is marked in insets as a red arrow and oriented towards the z-axis along the symmetry axis of the hybrid nanostructure with a certain distance G. Red solid line corresponds to G = 145 nm and green

700 800

Wavelength (nm)

Figure 4.2 — Calculated Purcell factor of gold-silicon nanoantenna with different gold shapes: top row corresponds to single disk-cone nanoantenna; middle row corresponds to single cup-cone nanoantenna; bottom row corresponds to single sphere-cone nanoantenna. The distance between the dipole source and gold-silicon nanoantenna is 145 nm for the red

solid line and 125 nm for the green dashed line

dashed line corresponds to G = 125 nm. The length of the dipole is set to be 20 nm which is less than A/15.

As discussed in section 1.3, the dipole interactions decay dramatically with the distance increasing. A similar effect is expected in such a case. As can be seen, a closer distance between the dipole source and gold-silicon nanoantenna contributes to a higher Purcell factor by comparing the green dashed line and the red solid line. And the distance between the dipole source and gold-silicon nanoantenna only influences the intensity of the Purcell factor without affecting the resonant wavelength.

On the other hand, some relation between the Purcell factor and scattering spectra is observed. For disk-cone nanoantenna, it is noticed that the same resonance around 910 nm appears in the Purcell factor spectrum and scattering spectrum. After the gold component shape changes from nanodisk to nanocup, the resonance in Purcell factor blue shifts to 700 nm, which is

also consistent with the scattering spectrum. This indicates the resonance in scattering and Purcell factor are related to each other. In order to design proper light-emitting systems, it is necessary to check the resonance in scattering at first. Besides, for the same wavelength around 700 nm, one can see the resonance is slightly enhanced and suppress the full width at half maximum (FWHM) after the gold component changes to nanosphere. This proves the tuning and enhanced effect of Purcell factor for laser- induced reshaping gold-silicon nanoantenna.

4.2 Near field control in dimers of gold-silicon nanoantennas

After analyzing a single gold-silicon nanoantenna at different reshaping stages to understand the mechanism, next step is to study pairs of goldsilicon nanoantenna separated by a subwavelength gap in the same excitation condition for three stages.

500 600 700 800 900 1000 Wavelength (nm)

Figure 4.3 — (a) Scattering spectra for pairs of gold-silicon nanoantenna at different fs-laser reshaping stages; (b) E-field distribution at the scattering resonance (670 nm)

The hybrid pair nanoantenna has the identical geometries for each one as the studied single gold-silicon nanoantenna and the distance between them

is set to be 100 nm. The scattering spectra and E-field distributions are calculated ans presented in Fig. 4.3. Comparing the scattering spectra for a single gold-silicon nanoantenna and a pair, they have comparable resonant behavior as well as the resonance shift during the shape modification from nanodisk to nanocup in hybrid pair nanoantenna. Here the resonance at 670 nm is marked for three modification stages. In Fig. 4.3b, Enhanced E-fields are presented due to the coupling between the individual gold-silicon nanoantenna. Particularly, after the nanodisks are modified to nanospheres, the field intensity in the gap is dramatically increased to value 2. It is predictable with a smaller distance between the pair gold-silicon nanoantenna, the field localization in the gap can be further enhanced due to stronger oscillation coupling.

700 800

Wavelength (nm)

Figure 4.4 — Calculated Purcell factor of a pair of gold-silicon nanoantenna with different

gold component shapes: top row corresponds to disk-cone nanoantenna; middle row corresponds to cup-cone nanoantenna; bottom row corresponds to sphere-cone nanoantenna. The distance between the dipole source and gold-silicon nanoantenna is 145 nm for the red

solid line and 125 nm for the green dashed line

Such electric field enhancement is supposed to induce Purcell factor

enhancement. Thus the Purcell factor of a pair gold-silicon nanoantenna with different shapes is calculated and plotted it in Fig. 4.4. Comparing the Purcell factor spectra for a single and a pair of gold-silicon nanoantenna, the resonance position is the same but the intensity achieved with pairs is twice higher than that of a single one. This is due to the near-field enhancement at the gap between the pair nanoantenna and thus induce the Purcell enhancement. Besides, the FWHM is mychn narrower than a single gold-silicon nanoantenna, indicating a high Q factor nanocavity, which can be promising for photoluminesence enhancement applications. To conclude, it is shown the shape and distance of the hybrid pair nanoantenna influence on the electric-field confinement inducing Purcell factor enhancement.

4.3 Chapter conclusions

This chapter reveals the shape and distance influence on the near-field control in single and pairs of gold-silicon nanoantenna. As LSPR depends on the shape and size of the plasmonic nanoparticle, the near-field confinement wavelength and intensity are modified with gold component reshaped in goldsilicon nanoantenna. Near-field tuning is a hot topic in nanophotonics but optical control in hybrid metal-dielectric nanostructures is still in progress. Thus this type of gold-silicon nanoantenna can be utilized to reconfigure and enhance the electric near-field distribution.

CHAPTER 5. RECONFIGURATION OF NEAR-FIELD DISTRIBUTION IN GOLD-SILICON HEPTAMER

Chapter 4 demonstrates the electric near-field enhancement and reconfiguration by controlling the shape and distance of gold-silicon nanoantennas. An identical size for the hybrid nanoantenna is chosen for lithography fabrication. Since near-field is polarization-dependent, heptamer consisted of seven gold-silicon nanoantennas is selected for experimental demonstration which is also the most popular oligomer configuration in published research works. In this case, obtained results of hybrid nanoantennas can be compared with both plasmonic nanostructures [29] and all-dielectric oligomers [31,103]. Researchers have proposed several experimental ways to tune the near-field distribution in plasmonic nanostructures, such as hot spots switching in azopoly-mer covered trimer gold nanorods by laser exposure [47], chemical reactions induced oxide layer for hot spot generation in native aluminium bowties [21]. The alternative proposed method lies in changing the excitation polarization of heptamers [29]. However, near-field tuning in hybrid metal-dielectric nanostructures is not reported yet. Thus in this chapter, the hybrid heptamer sample is fabricated and near-field reconfiguration as well as far-field modification experiments through fs-laser reshaping are performed. Section 5.1 represents the laser-induced reshape modification of gold components of goldsilicon heptamer, including the laser polarization influence on the scattering spectra and figure out the optimal way to map the near-field in hybrid hep-tamer. Section 5.2 shows the experimental SNOM mapping of the near-field of the hybrid gold-silicon heptamer at different fs-laser reshaping stages. Section 5.3 demonstrates the tuning effects of far-field properties of the hybrid heptamer as well as experimental and numerical modelling comparison.

5.1 Studies of laser-induced modification of gold-silicon heptamer

Fig. 5.1 schematically shows the configuration of the gold-silicon hep-tamer before and after fs-laser reshaping. Initially, the hybrid heptamer is

composed of seven gold-silicon nanoantenna which consists of Au nanodisks located on the truncated silicon nanocones (see Fig. 5.1 left part). After the fs-laser is applied to reshape the hybrid heptamer, the gold nanodisks are modified to gold nanocups (see Fig. 5.1 right part). In this case, the near-field distribution and scattering properties are also modified which is proved by the simulations in Chapter 4.

Figure 5.1 — Schematic view of gold-silicon heptamer before and after fs-laser reshaping

Thus the same geometry of gold-silicon nanoantenna is taken for sam-plw fabrication as in Chapter 4 which demonstrates the resonances in the visible range, isolated from the wavelength of fs-laser irradiation, i.e., 1050 nm. The hybrid heptamer is composed of six identical peripheral gold-silicon nanoantennnas sourrounding one relatively smaller hybrid nanoantenna. The bottom diameter and height of the peripheral silicon nanocones equal to the diameter of the gold nanodisks (190 nm) resulting from the etching process. The upper diameter of the peripheral nanocones are half of the bottom diameter (95 nm). The thickness of all the gold nanodisks are 20 nm. For the central hybrid nanoantenna, the silicon nanocone has a relatively smaller bottom diameter (150 nm) and upper diameter (60 nm). The distance between the central gold-silicon nanoantenna and the peripheral ones as well as between the outer hexagon is 200 nm, counting from center to center.

Fig. 5.2 shows the SEM image of the side view of hybrid heptamer before fs-laser reshaping. Since the near-field distribution is sensitive to the laser polarization and sample symmetry, it is necessary to start from measuring the sample with different polarization of the incident light and sample orientation. As can be seen from Fig. 5.2, the heptamer is symmetric along the x-axis and

Figure 5.2 — SEM image of the side view of the initial heptamer, scale bar is 200 nm

y-axis but not centrosymmetric. Thus the incident light along the x-axis is identified as horizontal excitation and along the y-axis is characterized as vertical excitation. Then the scattering spectra of the heptamer for horizontal and vertical sample orientation under TE- and TM-polarization are measured by confocal dark-field microscopy.

600 650 700 750 800 600 650 700 750 800

Wavelength(nm) Wavelength(nm)

Figure 5.3 — Dark-field spectra of the hybrid heptamer with horizontal and vertical orientation under (a) TE-polarization; (b) TM-polarization

As can be seen, despite the resonant wavelength is different from TE-and TM-polarization, the hybrid heptamer has the same resonance for both horizontal and vertical orientations under each polarization. This is also consistent with the numerical simulation in Chapter 4, when the oligomers have

more than four gold-silicon nanoantennas with a relatively smaller one in the center, the coupling between the central and peripheral nanoantennas is similar and contributes to comparable scattering spectra. Moreover, the resonant wavelength is not affected by the orientation and excitation polarization. In this case, further studies can only focus on the fs-laser modification of optical properties in one polarization. TE-polarization is chosen due to high noise ratio in experiments.

Next, the influence of laser polarization is investigated. A glan prism (Thorlabs) and a superachromatic quarter-wave plate (Thorlabs, 600-2700 nm) are adapted to construct horizontal linear polarization, vertical linear polarization and circular polarization of fs-laser to reshape the gold-silicon heptamer. The scattering spectra of the hybrid heptamer are measured before and after fs-laser reshaping for three polarization at the same laser flunence and plotted in Fig. 5.4. Comparing the normalized scattering spectra, one can see the comparable resonance shifts for different laser polarization. Detailed mode analysis of the resonance is demonstrated in section 5.3. It is assumed that the high rotation symmetry in the hybrid heptamer contributes to the neglecting of laser polarization influence. Thus it is possible to study the near-field reconfiguration during the process of fs-laser reshaping with simply linearly polarized laser excitation.

670 680 690 700 710 670 680 690 700 710 660 670 680 690 700 710

Wavelength, nm Wavelength, nm Wavelength, nm

Figure 5.4 — Normalized scattering spectra of the hybrid heptamer before and after different polarized laser modification: (a) horizontal linear polarization; (b) vertical linear

polarization; (c) circular polarization

5.2 Realization of near-field reconfiguration in hybrid heptamer

Despite the polarization dependence studied at the same laser fluence, near-field distribution is quite sensitive to the shapes and distances of plas-monics components. Therefore it is mandatory to investigate the near-field configuration of the hybrid heptamer with different reshaping stages of the gold components.

Initial nanostructure of the hybrid heptamer is consisted of gold nan-odisks and silicon nanocones. Fig. 5.5a represents the SEM image of the top view of initial heptamer. When the heptamer is exposed at 1.5 mJ/cm2 laser fluence, the shape of the nanodisk is just slightly modified comparing to the original shape. When the hybrid heptamer is irradiated by fs-laser at 3 mJ/cm2, Fig. 5.5b shows the first distinct reshaping of the central gold nanodisk due to its presence in the central position of the fs-laser beam and the relatively smaller size of the nanodisk compared to the peripheral nan-odisks. However, all the nanodisks are modified to nanocups when the hybrid heptamer is illuminated with the laser fluence at 4.5 mJ/cm2, as shown in Fig. 5.5c. As the melting temperature of gold (1337 K) is much lower than amorphous silicon (1687 K), and fs-laser irradiation induced heating at such fluence is between the melting temperature of gold and silicon, thus during the fs-laser reshaping process, only the shape of gold component is modified while the silicon nanocone remains the original configuration (see Fig. 7d, side view of Fig. 7c). Since higher laser fluences irradiation damages the well-defined configuration of the heptamer, the laser-induced modification experiments are limited to the fluence of 4.5 mJ/cm2. In this case, small modification in the hybrid heptamer is tracked and recorded, leading to better evaluation of the geometry changing influence on the near-field reconfiguration. For the near-field mapping experiments, only three cases are considered: the original hybrid heptamer, reshaped at 1.5 and 4.5 mJ/cm2.

An aperture-type near-field scanning optical microscope (AIST-NT, details see experimental characteristics in section 2.2.3) is utilized to visualize the laser-induced near-field reconfiguration. The hybrid heptamer sample is placed on a piezo stage and normally excited by a moderate focused laser

Figure 5.5 — SEM images of hybrid heptamers: top view for (a) non-irradiated hybrid nanoantennas; (c) and (d) after laser reshaping at 3 and 4.5 mJ/cm2 respectively; (d) side

view of (c). Scale bar is 200 nm

beam (Fianinum WhiteLase Supercontinnum) which can not induce any shape modification of the hybrid nanoantennas from the substrate side. The laser beam is linearly polarized along the y-axis (see Fig. 5.5). In order to inhibit the signal artifacts ascribed to the sample's complex topography and possible damage of the heptamer in the process of scanning, these near-field images are recorded in a constant height mode (details see section 2.2.3). Here the laser source as well as the sample are fixed and the probe aperture is moved within a plane 120 nm on the top of the gold components of the initial hybrid heptamer. The near-field configuration is measured with 20 nm step size within the range of 640-740 nm of three cases (initial, 1.5 and 4.5 mJ/cm2 laser irradiation), as plotted in Fig. 5.6.

Fig. 5.6 represents the strong modified and enhanced near-field distri-

Figure 5.6 — SNOM measurements of hybrid heptamer reshaped at different laser fluences: (a) initial heptamer with nanodisks; (b) 1.5 mJ/cm2 with moderate modification; (c) 4.5 mJ/cm2 with nanocups. The dimension of the near-field map is 3pm x 3pm

butions of the heptamer causing by fs-laser reshaping. Insets in the 640 nm column emphasize the position and configuration of the hybrid heptamer. The most pronounced modification locates in the range of 680-740 nm. As can be seen from the upper row (Fig. 5.6a), it only shows a very weak field concentration spot above the center of the heptamer. However, when 4.5 mJ/cm2 laser fluence is applied for laser reshaping, the field spot remains the same position in the center and the intensity increases around two times than the initial nanostructure (see Fig. 5.6c). Moreover, with a moderate reshaping at 1.5 mJ/cm2, the single hot spot changes to four strongly enhanced hot spots around the slightly modified central gold nanodisk (see Fig. 5.6b). To clearly showing the field intensity variation, the maximum SNOM intensity of the hot spots in the center for each map is depicted in Fig. 5.7.

The hot spots in the hybrid heptamer are very sensitive to the shape and distance between the gold nanodisks. This can be precisely controlled by adjusting the irradiation laser fluence. With the increasing of the laser fluence, the central nanodisk possessing a smaller diameter and located in the center of the laser beam starts to melt and modify the shape to nanocup. For laser fluence at 1.5 mJ/cm2, the distance between the central nanodisk and peripheral nanodisks is larger than the distance between the peripheral nan-

2

1.8

1.6

1.4

1.2

(fl

c

<D 1

C

jtf O Initial

0,8 Jtd--

/ * 1.5 mJ/cm2 0.6 - -

* 4.5 mJ/cm 2 0.4 ■ -

0.2 ■

_i_i_i_i_i_■

640 660 680 700 720 740

Wavelength (rim)

Figure 5.7 — Extracting maximum SNOM signal in the center of the hybrid heptamer

reshaped at different laser fluences: blue circle identifies the initial heptamer with nanodisks; green star indicates 1.5 mJ/cm2 with moderate modification of the central nanodisk; red cross refers to 4.5 mJ/cm2 with nanocups

odisks, resulting in the transformation of hot spot numbers and distributions in Fig. 5.6b. This is so-called distance-related bonding and anti-bonding modes in oligomers which also demonstrates in [14,29,30]. In this case, reconfiguration of the near-field in hybrid heptamer can be achieved by fs-laser induced shape variation of gold nanodisks after lithography fabrication. It should be mentioned that for plasmonic nanostructures, aperture-type SNOM measurements possess strong tip-sample interactions, thus it is very challenging to directly compare the measured near-field with the numerically calculated near-field quantitatively [104].

5.3 Far-field reconfiguration studies

In contrast to the dramatic modifications in the near-field distribution, the far-field properties of the hybrid heptamer are not strongly changed. Here Fig. 5.8a shows the scattering spectra before and after reshaping to cups